---
title: "Through the Grapevine: Informational Consequences of Interpersonal Political Communication"
subtitle: "Online Appendix"
author: "Taylor N. Carlson"
geometry: margin=2cm
header-includes:
  - \renewcommand{\thetable}{\arabic{table}}%
  - \renewcommand{\thefigure}{\arabic{figure}}%
  - \usepackage{booktabs}
  - \usepackage{dcolumn}
  - \usepackage{pdflscape}
  - \usepackage{epsf,graphicx,psfrag}
  - \usepackage{natbib}
  - \usepackage{setspace}
  - \usepackage{fullpage}
  - \usepackage{endnotes}
  - \usepackage{natbib}
  - \usepackage{amsfonts}
  - \usepackage{amssymb,enumerate}
  - \usepackage{dcolumn}
  - \usepackage{amssymb}
  - \usepackage{nicefrac}
  - \usepackage{pdfsync}
  - \usepackage{url}
  - \usepackage{pst-all}
  - \linespread{1.05} 
  - \usepackage{placeins}
  - \usepackage{afterpage} 
  - \usepackage{ragged2e}
  - \usepackage{multirow}
  - \usepackage{multicol}
  - \usepackage{fancyhdr}
  - \renewcommand{\headrulewidth}{0pt}
  - \usepackage{subfig}
  - \renewcommand{\figurename}{Fig.}
  - \usepackage[labelfont=bf]{caption}
  - \captionsetup[figure]{labelsep=space}
  - \usepackage{array}
output:
  pdf_document:
    fig_caption: yes
    keep_tex: yes
    toc: yes
    toc_depth: 3
  html_document:
    fig_caption: yes
    highlight: pygments
    keep_md: no
    lib_dir: libs
    mathjax: null
    template: default.html
    theme: default
    toc: no
---

```{r global_options, include=FALSE, cache=FALSE}
rm(list=ls())
knitr::opts_chunk$set(fig.width=6, fig.height=6, fig.path='figures/', warning=FALSE, message=FALSE, fig.retina=NULL, cache=F, autodep=T, echo=FALSE)
library(kfigr) #figure referencing for markdown
library(stargazer) #pretty tables
library(xtable) #pretty tables
library(knitr) #knitting
```

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r data, echo=FALSE, eval=TRUE}
rm(list=ls())
# Set working directory
setwd("/Users/Taylor/Dropbox/Consequences of Social Information/APSR Submission/APSR R&R/APSR Replication Materials/Carlson_Grapevine_Replication_Materials/")

```

# Study 1

```{r echo=FALSE, eval=TRUE}
# Read in study 1 dataset

s1 <- read.csv("grapevine_study1_121318.csv")
s1 <- s1[-1,] # Remove first row, which just contains the news article information

```

## Sample Characteristics

```{r study1_sample, echo=FALSE, eval=TRUE, results="asis"}
# Create a demographics table for study 1
## party ID (dem, rep, ind), ideology, gender, race, political interest

s1$dem.sender <- NA
s1$dem.sender[s1$pid7.sender<4] <- 1
s1$dem.sender[s1$pid7.sender>=4] <- 0

s1$rep.sender <- NA
s1$rep.sender[s1$pid7.sender>4] <- 1
s1$rep.sender[s1$pid7.sender<=4] <- 0

s1$ind.sender <- NA
s1$ind.sender[s1$pid7.sender==4] <- 1
s1$ind.sender[s1$pid7.sender!=4] <- 0

s1$white.sender <- NA
s1$white.sender[s1$race.sender==1] <- 1
s1$white.sender[s1$race.sender!=1] <- 0

s1$copartisan.sender <- NA
s1$copartisan.sender[s1$pid7<4 & s1$pid7.sender<4] <-1
s1$copartisan.sender[s1$pid7==4 & s1$pid7.sender==4] <- 1
s1$copartisan.sender[s1$pid7>4 & s1$pid7.sender>4] <-1
s1$copartisan.sender[s1$pid7<4 & s1$pid7.sender>=4] <-0
s1$copartisan.sender[s1$pid7>4 & s1$pid7.sender<=4] <-0
s1$copartisan.sender[s1$pid7==4 & s1$pid7.sender!=4] <-1
#table(s1$copartisan.sender)


study1sample <- matrix(NA, nrow=7, ncol=2)

study1sample[1,1:2] <- c(mean(s1$rep.sender, na.rm=T), table(s1$rep.sender)[2])
study1sample[2,1:2] <- c(mean(s1$dem.sender, na.rm=T), table(s1$dem.sender)[2])
study1sample[3,1:2] <- c(mean(s1$ind.sender, na.rm=T), table(s1$ind.sender)[2])
study1sample[4,1:2] <- c(mean(s1$female.sender, na.rm=T), table(s1$female.sender)[2])
study1sample[5,1:2] <- c(mean(s1$white.sender, na.rm=T), table(s1$white.sender)[2])
study1sample[6,1:2] <- c(mean(s1$ideology.sender, na.rm=T), sum(table(s1$ideology.sender)))
study1sample[7,1:2] <- c(mean(s1$interest.sender_r, na.rm=T), sum(table(s1$interest.sender_r)))

rownames(study1sample) <- c("Republican", "Democrat", "Independent", "Female", "White", "Ideology (1=extremely liberal, 7=extremely conservative)", "Political Interest (1=not at all interested, 4=very interested")
colnames(study1sample) <- c("Prop/Mean", "Number")

options(xtable.comment=FALSE)
print(xtable(study1sample, caption="Descriptive Statistics for Study 1 Mechanical Turk Sample", type="latex"),
      caption.placement="top", size="scriptsize")
```

\FloatBarrier

The first two columns show the sample characteristics for the full sample in Study 1.  The third and fourth columns show the sample characteristics for the participants who generated messages that got passed on in the Study 2 experiment. Overall, the Mechanical Turk sample was dominated by Democrats, with about `r round(study1sample[2,1]*100, digits=2)`\% of respondents identifying as Democrats, `r round(study1sample[1,1]*100, digits=2)`\% identifying as Republicans, and `r round(study1sample[3,1]*100, digits=2)`\% identifying as Independents. The sample was evenly split between men and women, but was dominated by Whites, with only `r 100-round(study1sample[5,1]*100, digits=2)`\% of the sample identifying as non-white. This sample is surely not nationally representative. 

## Balance Tables

```{r study1balance, echo=FALSE, eval=TRUE, results="asis"}

s1$study1_treatment <- NA
s1$study1_treatment[s1$pid7<4] <- "To Democrat"
s1$study1_treatment[s1$pid7==4] <- "To Independent"
s1$study1_treatment[s1$pid7>4] <- "To Republican"

demogs <- c("rep.sender", "dem.sender", "ind.sender", "female.sender", "white.sender", "ideology.sender", "interest.sender_r")
study1balance <- matrix(NA, nrow=length(demogs), ncol=3)
for(i in 1:nrow(study1balance)){
  study1balance[i,] <- c(mean(s1[, demogs[i]][s1$study1_treatment=="To Democrat"], na.rm=T), #Democrat recipient
                         mean(s1[, demogs[i]][s1$study1_treatment=="To Republican"], na.rm=T), # Republican recipient 
                         mean(s1[, demogs[i]][s1$study1_treatment=="To Independent"], na.rm=T)) #Independent recipient
}
rownames(study1balance) <- c("Republican", "Democrat", "Independent", "Female", "White", "Ideology (1=extremely liberal, 7=extremely conservative)", "Political Interest (1=not at all interested, 4=very interested")
colnames(study1balance) <- c("To Democrat", "To Republican", "To Independent")

options(xtable.comment=FALSE)
print(xtable(study1balance, caption="Balance on Observables between Treatment Groups", type="latex"),
      caption.placement="top")


```

\FloatBarrier

Looking at balance between treatment groups, we can see that overall the groups were fairly balanced. There were more women randomly assigned to write messages to Independents than to Republicans. Messages written to Independents were generated by individuals with slightly lower levels of political interest than messages written to Democrats or Republicans. 

## Survey Instrument 

### Consent

[NAME REDACTED] is conducting a research study to find out more about how individuals learn about politics. You have been asked to take part because you are a member of Mechanical Turk and have indicated an interest in participating. The purpose of this study is to gether data on your responses to a short series of survey questions. 

If you agree to be in this study, the following will happen to you:

\begin{enumerate}
  \item To complete the study you will complete an online survey on a computer of your choice 
  \item While taking the survey, you will complete survey questions online.
\end{enumerate}

Participation in this study may involve some added risks or discomforts. The main risks associated with these procedures are fatigue, irritability, and frustration with the testing procedure. Because this is a research study, there may also be some unknown risks that are currently unforeseeable. You will be told if any important new information is found during the course of this study that may affect your wanting to continue. 

There will not be any direct benefit to you from participating in this study. Society may benefit from the knowledge about political information acquired by the investigator in this study. 

Participation in research is entirely voluntary. You may refuse to participate or withdraw or refuse to answer specific questions in an interview or on a questionnaire at any time. The alternative to participation in this study is not to participate. 

Your responses on this survey will be strictly anonymous. No identifying information will be collected. All data collected will be stored in password-protected files that will only be accessible by the researchers. Some written responses may be used to inform future surveys, but no identifying information will be kept or used with them. If you have any additional questions about the steps we have taken to assure your privacy please do not hesitate to ask by e-mail. 

If completed on Mechanical Turk, you will be compensated with \$0.50 for your participation, which equates to a rate of approximately \$7.00 per hour. There will be no cost to you for participating in this study. 

If you have other questions or research-related problems, you may reach [NAME REDACTED] at [CONTACT INFORMATION REDACTED]. You may call the Human Research Protections Program Office at [CONTACT INFORMATION REDACTED] to inquire about your rights as a research subject or to report research-related problems. 

By completing and submitting the survey you confirm that you are at least 18 years of age and consent to participate in this study. 

\begin{itemize}
  \item I consent to participate in this study
  \item I do not consent to participate in this study
\end{itemize}

### Survey Questions 

\begin{itemize}
  \item Please spend a few minutes reading the following article about U.S. economic performance in the first quarter of 2017. You can spend as much time reading it as you like, but we ask that you read it as if you were trying to learn about the econoy or read the news in your daily life. 
  \item PAGE BREAK
  \item Shown News Article 
\end{itemize}

\textbf{U.S. first-quarter GDP growth revised up to 1.4 percent} 
By Lindsay Dunsmuir WASHINGTON, JUNE 29 Reuters

The U.S. economy slowed less sharply in the first quarter than initially estimated due to unexpectedly higher consumer spending and a bigger jump in exports.

Gross domestic product increased at a 1.4 percent annual rate instead of the 1.2 percent pace reported last month, the Commerce Department said in its final assessment on Thursday.

It was still the slowest growth rate since the second quarter of last year. Economists polled by Reuters had expected GDP growth to remain unchanged at a 1.2 percent rate.

GDP for the January-March period tends to underperform relative to the rest of the year due to perennial issues with the calculation of the data the government has said it is working to resolve.

First-quarter economic growth was boosted by an upward revision to consumer spending, which accounts for more than two-thirds of U.S. economic activity. Consumer spending rose at a 1.1 percent rate instead of the previously reported 0.6 percent pace. It was still the slowest pace since the second quarter of 2013.

Despite the upward revision, the Trump administration's stated target of swiftly boosting U.S. growth to 3 percent remains a challenge.

A sustained average of 3 percent growth has not been seen since the 1990s. Since 2000, the U.S. economy has grown at an average 2 percent rate. The U.S. economy expanded 1.6 percent in 2016, the lowest rate in five years.

President Donald Trump's economic program of tax cuts, regulatory rollbacks and infrastructure spending has yet to get off the ground five months into his presidency.

Initial signs that economic growth re-accelerated sharply in the second quarter have also faltered with recent disappointing data on retail sales, manufacturing production and inflation. Housing data has also been mixed. The Atlanta Federal Reserve currently forecasts annualized GDP growth of 2.9 percent in the second quarter.

Exports in the first quarter were revised to show a gain of 7.0 percent from the previously reported 5.8 percent.

Business spending on equipment was revised to show it increasing at a 7.8 percent rate in the January-March period rather than the 7.2 percent previously estimated.

Businesses accumulated inventories at a rate of \$2.6 billion in the first quarter, rather than the \$4.3 billion reported last month. Inventory investment rose at a \$49.6 billion rate in the fourth quarter of last year.

Inventories subtracted 1.11 percentage point from GDP growth instead of the 1.07 percentage point previously reported.

The government also reported that corporate profits after tax with inventory valuation and capital consumption adjustments fell at an annual rate of 2.7 percent in the first quarter after rising at a 2.3 percent pace in the prior three months.

\begin{itemize}
  \item PAGE BREAK
  \item RANDOMLY ASSIGNED TO ONE OF THE FOLLOWING PROMPTS WITH A LARGE TEXT BOX
    \begin{itemize}
    \item Imagine that you were discussing politics and current events with a Republican.  Please write what you would tell a Republican about the article you just read. Please do not include any names or identifying information about you or the people you know.
    \item Imagine that you were discussing politics and current events with a Democrat.  Please write what you would tell a Democrat about the article you just read. Please do not include any names or identifying information about you or the people you know.
    \item Imagine that you were discussing politics and current events with an Independent.  Please write what you would tell an Independent about the article you just read. Please do not include any names or identifying information about you or the people you know.
    \end{itemize}
  \item PAGE BREAK
  \item Thank you for writing your message. There are a few more short questions on this survey.
  \item PAGE BREAK
  \item How would you rate the political bias, if any, of the article you just read?
    \begin{itemize}
    \item Strongly favored Democrats
    \item Favored Democrats
    \item Slightly favored Democrats
    \item No bias
    \item Slightly favored Republicans
    \item Favored Republicans
    \item Strongly favored Republicans
    \end{itemize}
  \item Before taking this survey, had you read the article we presented you with before?
    \begin{itemize}
    \item Yes
    \item Not that exact article, but I've read others similar to it
    \item No
    \end{itemize}
  \item How much do you trust the content of this article to be accurate and reliable?
    \begin{itemize}
    \item A great deal
    \item A lot
    \item A moderate amount 
    \item A little
    \item None at all
    \end{itemize}
  \item Which of the following best describes what you were trying to accomplish with the message you wrote to another person about the US economic performance?
    \begin{itemize}
    \item I was trying to objectively inform the other person
    \item I was trying to persuade the other person to view the economy or politicans the way I do 
    \item I was trying to convince the other person to get involved in politics 
    \item I was trying to mislead the other person about the state of the economy
    \end{itemize}
  \item PAGE BREAK
  \item We'd like to ask you a few questions about the current state of the US economy. Please indicate whether you think each of the following statements are TRUE or FALSE.
    \begin{itemize}
    \item The US economy grew in the first quarter of 2017
    \item In the first quarter of 2017, GDP grew at the fastest rate since the second quarter of 2016
    \item Consumer spending accounts for less than 1/4 of US economic activity 
    \item GDP in January-March tends to over-perform relative to the rest of the year 
    \item Since 2000, the US economy has grown at an average rate of 0.5\%
    \item In the first quarter of 2017, GDP grew at a slower rate than the Trump administration's target
    \end{itemize}
  \item PAGE BREAK
  \item Below are your responses to the previous questions. How confident are you that you answered each question correctly? [Each statement shown, followed by the answer (True or False) selected by the participant]
    \begin{itemize}
    \item Not very confident
    \item Somewhat confident
    \item Confident
    \item Very confident 
    \item Don't Know
    \end{itemize}
  \item PAGE BREAK
  \item Right now, do you think that economic conditions in the country as a whole are getting better or getting worse?
    \begin{itemize}
    \item Getting better
    \item Getting worse
    \item Staying the same
    \item Don't know
    \end{itemize}
  \item How would you rate the economic conditions in this country today?
    \begin{itemize}
    \item Excellent
    \item Good
    \item Only fair
    \item Poor
    \item Don't know
    \end{itemize}
  \item PAGE BREAK
  \item Do you approve or disapprove of the way Donald Trump is handling his job as president?
    \begin{itemize}
    \item Strongly approve
    \item Approve
    \item Disapprove
    \item Strongly disapprove
    \item No Opinion 
    \end{itemize}
  \item Do you approve or disapprove of the way Donald Trump is handling the economy?
    \begin{itemize}
    \item Strongly 
    \item Approve
    \item Disapprove
    \item Strongly disapprove
    \item No Opinion 
    \end{itemize}
  \item PAGE BREAK
  \item When thinking politically, what do you usually consider yourself---a Republican, a Democrat, an Independent, or what?
    \begin{itemize}
    \item Democrat
    \item Republican
    \item Independent
    \end{itemize}
  \item PAGE BREAK
  \item IF INDEPENDENT: Do you lean more towards the Republican Party or the Democratic Party?
      \begin{itemize}
      \item Republican
      \item Democratic
      \item Neither
      \end{itemize}
  \item IF REPUBLICAN OR DEMOCRAT: Do you consider yourself a strong [REPUBLICAN/DEMOCRAT piped in] or a not very strong [REPUBLICAN/DEMOCRAT piped in]?
    \begin{itemize}
    \item Strong [REPUBLICAN/DEMOCRAT piped in]
    \item Not a very strong [REPUBLICAN/DEMOCRAT piped in]
    \end{itemize}
  \item PAGE BREAK
  \item Would you like to learn more about the US economic performance? If so, please select which articles you would be interested in reading and we will provide you with links to those articles.
    \begin{itemize}
    \item US economic growth in Q1 upgraded to 1.4 percent (Fox News, Associated Press)
    \item US economic growth in Q1 upgraded to 1.4\% (USA Today, Associated Press)
    \item Gross Domestic Product: First Quarter 2017 (Third Estimate) (Bureau of Economic Analysis)
    \item Final reading on Q1 gross domestic product up 1.4\% vs 1.2\% rise expected (CNBC, Reuters)
    \item U.S. GDP Growth Revised Up to 1.4\% Rate in First Quarter (Fox Business)
    \item First Quarter GDP Growth Bumped Up to 1.4 Percent (US News and World Report)
    \item The Trump economy got off to a sluggish start (CNN Money)
    \end{itemize}
  \item PAGE BREAK
  \item Here are the links from the previous page. You may view them if you would like or open them in a new window to view after completing this survey. [selected links to articles shown below]
  \item PAGE BREAK 
  \item How likely are you to do each of the following? [Very unlikely, Unlikely, Likely, Very likely, Don't Know]
    \begin{itemize}
    \item Discuss the US economy with someone else
    \item Post a status on Facebook or a Tweet about the US economy
    \item Research more informatin on the US economy 
    \item Research more information on President Trump's economic policies 
    \item Contact your representative about the US economy 
    \item Subscribe to newsletters from the Bureau of Economic Analysis 
    \item Donate to President Trump's reelection campaign 
    \item Donate to a candidate seeking to challenge President Trump 
    \end{itemize}
  \item PAGE BREAK
  \item When it comes to politics, would you describe yourself as liberal, conservative, or neither liberal nor conservative?
    \begin{itemize}
    \item Very liberal
    \item Somewhat liberal 
    \item Closer to liberals 
    \item Neither liberal nor conservative 
    \item Closer to conservatives
    \item Somewaht conservative 
    \item Very conservative 
    \end{itemize}
  \item PAGE BREAK
  \item With which gender do you identify?
    \begin{itemize}
    \item Male
    \item Female
    \end{itemize}
  \item With which racial or ethnic group do you most strongly identify?
    \begin{itemize}
    \item White
    \item Black or African American
    \item American Indian or Alaska Native
    \item Asian
    \item Native Hawaiian or Pacific Islander
    \item Latino or Hispanic
    \item Other
    \end{itemize}
  \item PAGE BREAK
  \item How interested are you in politics and public affairs?
    \begin{itemize}
    \item Very interested
    \item Somewhat interested
    \item Not very interested
    \item Not at all interseted
    \end{itemize}
  \item PAGE BREAK
  \item How often do people ask you for information about politics and current events?
    \begin{itemize}
    \item Daily 
    \item 4-6 times per week
    \item 2-3 times per week
    \item Once a week
    \item Once a month 
    \item Twice a year
    \item Once a year
    \item Never
    \end{itemize}
\end{itemize}

\FloatBarrier

## Amount of Information Communicated

### Word Count by Partisanship of Sender and Receiver

```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=7, fig.cap="Word Count in Social Messages by Partisanship. These differences are not statistically significant."}

news <- read.csv("grapevine_news.csv")
soc <- read.csv("grapevine_study1_121318.csv")

soc$copartisan.sender <- NA
soc$copartisan.sender[soc$pid7<4 & soc$pid7.sender<4] <-1
soc$copartisan.sender[soc$pid7==4 & soc$pid7.sender==4] <- 1
soc$copartisan.sender[soc$pid7>4 & soc$pid7.sender>4] <-1
soc$copartisan.sender[soc$pid7<4 & soc$pid7.sender>=4] <-0
soc$copartisan.sender[soc$pid7>4 & soc$pid7.sender<=4] <-0
soc$copartisan.sender[soc$pid7==4 & soc$pid7.sender!=4] <-1
table(soc$copartisan.sender)

# Length

wordcountsoc <- matrix(NA, nrow=3, ncol=3)
wordcountsoc[1,] <- c(mean(s1$wordcount[s1$pid7.sender<4 & s1$study1_treatment=="To Democrat"], na.rm=T),
                      mean(s1$wordcount[s1$pid7.sender>4 & s1$study1_treatment=="To Democrat"], na.rm=T),
                      mean(s1$wordcount[s1$pid7.sender==4 & s1$study1_treatment=="To Democrat"],na.rm=T))

wordcountsoc[2,] <- c(mean(s1$wordcount[s1$pid7.sender<4 & s1$study1_treatment=="To Republican"], na.rm=T),
                      mean(s1$wordcount[s1$pid7.sender>4 & s1$study1_treatment=="To Republican"], na.rm=T),
                      mean(s1$wordcount[s1$pid7.sender==4 & s1$study1_treatment=="To Republican"],na.rm=T))

wordcountsoc[3,] <- c(mean(s1$wordcount[s1$pid7.sender<4 & s1$study1_treatment=="To Independent"], na.rm=T),
                      mean(s1$wordcount[s1$pid7.sender>4 & s1$study1_treatment=="To Independent"], na.rm=T),
                      mean(s1$wordcount[s1$pid7.sender==4 & s1$study1_treatment=="To Independent"],na.rm=T))

wordcountsoc.se <- matrix(NA, nrow=3, ncol=3)
#To Democrats
wordcountsoc.se[1,] <- c(sd(s1$wordcount[s1$pid7.sender<4 & s1$study1_treatment=="To Democrat"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Democrat"], s1$pid7.sender[s1$study1_treatment=="To Democrat"])[,1:3])),
                      sd(s1$wordcount[s1$pid7.sender>4 & s1$study1_treatment=="To Democrat"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Democrat"], s1$pid7.sender[s1$study1_treatment=="To Democrat"])[,5:7])),
                      sd(s1$wordcount[s1$pid7.sender==4 & s1$study1_treatment=="To Democrat"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Democrat"], s1$pid7.sender[s1$study1_treatment=="To Democrat"])[,4])))

#To Republicans
wordcountsoc.se[2,] <- c(sd(s1$wordcount[s1$pid7.sender<4 & s1$study1_treatment=="To Republican"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Republican"], s1$pid7.sender[s1$study1_treatment=="To Republican"])[,1:3])),
                      sd(s1$wordcount[s1$pid7.sender>4 & s1$study1_treatment=="To Republican"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Republican"], s1$pid7.sender[s1$study1_treatment=="To Republican"])[,5:7])),
                      sd(s1$wordcount[s1$pid7.sender==4 & s1$study1_treatment=="To Republican"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Republican"], s1$pid7.sender[s1$study1_treatment=="To Republican"])[,4])))

#To Independents
wordcountsoc.se[3,] <- c(sd(s1$wordcount[s1$pid7.sender<4 & s1$study1_treatment=="To Independent"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Independent"], s1$pid7.sender[s1$study1_treatment=="To Independent"])[,1:3])),
                      sd(s1$wordcount[s1$pid7.sender>4 & s1$study1_treatment=="To Independent"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Independent"], s1$pid7.sender[s1$study1_treatment=="To Independent"])[,5:7])),
                      sd(s1$wordcount[s1$pid7.sender==4 & s1$study1_treatment=="To Independent"], na.rm=T)/sqrt(sum(table(s1$wordcount[s1$study1_treatment=="To Independent"], s1$pid7.sender[s1$study1_treatment=="To Independent"])[,4])))

barCenters <- barplot(wordcountsoc, beside=T, main="Word Count in Social Messages\nby Partisanship",
                      ylim=c(0,60), ylab="Number of Words",
                      names.arg=c("From\nDemocrat", "From\nRepublican", "From\nIndependent"),
                      col=c("dodgerblue2", "firebrick2", "seashell4"))
legend(1,50, legend=c("To Democrat", "To Republican", "To Independent"), cex=.75, 
       fill=c("dodgerblue2", "firebrick2", "seashell4"), bty="n")
segments(barCenters, wordcountsoc - wordcountsoc.se*1.96, barCenters, wordcountsoc + wordcountsoc.se*1.96)

```

\FloatBarrier

## Hand-Coded Units of Information

### Coding Parameters

In addition to using word count as a rough proxy for the amount of information communicated, I examine how many units of information were included in each message (Moussaid et al. 2015; Carlson 2018). A unit of information is a single identifiable piece of information. Two independent coders read each message and enumerated the units of information contained in each message. The coders followed the coding scheme introduced by Moussaid et al. 2015. The coders obtained reasonable levels of inter-coder reliability for the total units of information contained in each message (Krippendorff's Alpha=.776; correlation=.82).

\begin{table}[h] \centering \footnotesize
\caption{Example Socially Transmitted Messages and Units of Information Coding}
\label{units}
\begin{tabular}{lll}
\hline
\parbox[t]{2.25in}{Message} & \parbox[t]{2.25in}{Units of Information} & \parbox[t]{1.25in}{Total Units} \\ \hline
\parbox[t]{2.25in}{All the things you feared would happen with trump are happening, and despite all the promises he made, he has not come through with any of them yet, and the outlook for keeping his promises looks NOT promising.} & \parbox[t]{2.25in}{1. All the things you feared would happen with Trump are happening; 2. Despite all the promises he made, he has not come through with any of them yet; 3. The outlook for keeping his promises looks NOT promising} & \parbox[t]{1.25in}{3} \\ \hline
\parbox[t]{2.25in}{Although the economy is not improving at the rate that it has in the recent past, this past quarter's performance was better than projected.} & \parbox[t]{2.25in}{1. The economy is not improving at the rate that it has in the recent past; 2. This past quarter's performance was better than projected} & \parbox[t]{1.25in}{2} \\ \hline
\parbox[t]{2.25in}{Basically Trump is growing America with the economy and jobs which trickle into many other thing like safety and security.} & \parbox[t]{2.25in}{1. Trump is growing America with the economy and jobs; 2. The economy is doing well; 3. Jobs are being created; 4. Trump improving the economy also improves safety and security} & \parbox[t]{1.25in}{4} \\ \hline
\end{tabular}
\end{table}

\FloatBarrier

### Total Units of Information

The average number of units of information in each message, according to Coder 1 was `r round(mean(s1$units.total1, na.rm=T), digits=2)` and according to Coder 2 was `r round(mean(s1$units.total2, na.rm=T), digits=2)`. The two coders correlated at r=`r round(cor.test(s1$units.total1, s1$units.total2)$estimate, digits=2)` (p<.001) and had a Krippendorff's Alpha above .7. The tables that follow show the units of information transmitted between each combination of sender-receiver partisanship. The first table shows the average of the two coders, followed by Coder 1, then Coder 2. While the numbers are slightly different, the patterns are the same across the coders. 

```{r echo=FALSE, eval=TRUE, results="asis"}
# Create a table showing the total units of information by coder and on average 
library(xtable)

total.tab <- matrix(NA, nrow=3, ncol=3)

## Start filling the table with the average number of units in each combination of sender/receiver partisanship

### Row 1, columns 1-3
total.tab[1,] <- c(mean(s1$units.total[s1$pid7.sender>4 & s1$pid7>4], na.rm=T), # party of sender is greater than 4 (Rep), party of receiver is greater than 4 (Rep)
                   mean(s1$units.total[s1$pid7.sender>4 & s1$pid7==4], na.rm=T),# party of sender is > 4 (rep), party of receiver is 4 (Ind)
                   mean(s1$units.total[s1$pid7.sender>4 & s1$pid7<4], na.rm=T)) # party of sender is > 4 (rep), party of receiver is <4 (dem)

### Row 2, columns 1-3
total.tab[2,] <- c(mean(s1$units.total[s1$pid7.sender==4 & s1$pid7>4], na.rm=T), # party of sender ==4 (ind), party of receiver >4 (rep)
                   mean(s1$units.total[s1$pid7.sender==4 & s1$pid7==4], na.rm=T),# party of sender is == 4 (ind), party of receiver is 4 (Ind)
                   mean(s1$units.total[s1$pid7.sender==4 & s1$pid7<4], na.rm=T)) # party of sender is ==4 (ind), party of receiver is <4 (dem)

### Row 3, columns 1-3
total.tab[3,] <- c(mean(s1$units.total[s1$pid7.sender<4 & s1$pid7>4], na.rm=T), # party of sender <4 (dem), party of receiver >4 (rep)
                   mean(s1$units.total[s1$pid7.sender<4 & s1$pid7==4], na.rm=T),# party of sender is < 4 (dem), party of receiver is 4 (Ind)
                   mean(s1$units.total[s1$pid7.sender<4 & s1$pid7<4], na.rm=T)) # party of sender is < 4 (dem), party of receiver is <4 (dem)

colnames(total.tab) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(total.tab) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=FALSE)
print(xtable(total.tab, caption="Units of Information by Sender and Receiver Partisanship (Average of Two Coders)", type="latex"), caption.placement="top")

```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, results="asis"}
# Create a table showing the total units of information by coder and on average 

total.tab1 <- matrix(NA, nrow=3, ncol=3)

## Start filling the table with the average number of units in each combination of sender/receiver partisanship

### Row 1, columns 1-3
total.tab1[1,] <- c(mean(s1$units.total1[s1$pid7.sender>4 & s1$pid7>4], na.rm=T), # party of sender is greater than 4 (Rep), party of receiver is greater than 4 (Rep)
                   mean(s1$units.total1[s1$pid7.sender>4 & s1$pid7==4], na.rm=T),# party of sender is > 4 (rep), party of receiver is 4 (Ind)
                   mean(s1$units.total1[s1$pid7.sender>4 & s1$pid7<4], na.rm=T)) # party of sender is > 4 (rep), party of receiver is <4 (dem)

### Row 2, columns 1-3
total.tab1[2,] <- c(mean(s1$units.total1[s1$pid7.sender==4 & s1$pid7>4], na.rm=T), # party of sender ==4 (ind), party of receiver >4 (rep)
                   mean(s1$units.total1[s1$pid7.sender==4 & s1$pid7==4], na.rm=T),# party of sender is == 4 (ind), party of receiver is 4 (Ind)
                   mean(s1$units.total1[s1$pid7.sender==4 & s1$pid7<4], na.rm=T)) # party of sender is ==4 (ind), party of receiver is <4 (dem)

### Row 3, columns 1-3
total.tab1[3,] <- c(mean(s1$units.total1[s1$pid7.sender<4 & s1$pid7>4], na.rm=T), # party of sender <4 (dem), party of receiver >4 (rep)
                   mean(s1$units.total1[s1$pid7.sender<4 & s1$pid7==4], na.rm=T),# party of sender is < 4 (dem), party of receiver is 4 (Ind)
                   mean(s1$units.total1[s1$pid7.sender<4 & s1$pid7<4], na.rm=T)) # party of sender is < 4 (dem), party of receiver is <4 (dem)

colnames(total.tab1) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(total.tab1) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=FALSE)
print(xtable(total.tab1, caption="Units of Information by Sender and Receiver Partisanship (Coder 1)", type="latex"), caption.placement="top")

```

\FloatBarrier


```{r echo=FALSE, eval=TRUE, results="asis"}
# Create a table showing the total units of information by coder and on average 

total.tab2 <- matrix(NA, nrow=3, ncol=3)

## Start filling the table with the average number of units in each combination of sender/receiver partisanship

### Row 1, columns 1-3
total.tab2[1,] <- c(mean(s1$units.total2[s1$pid7.sender>4 & s1$pid7>4], na.rm=T), # party of sender is greater than 4 (Rep), party of receiver is greater than 4 (Rep)
                   mean(s1$units.total2[s1$pid7.sender>4 & s1$pid7==4], na.rm=T),# party of sender is > 4 (rep), party of receiver is 4 (Ind)
                   mean(s1$units.total2[s1$pid7.sender>4 & s1$pid7<4], na.rm=T)) # party of sender is > 4 (rep), party of receiver is <4 (dem)

### Row 2, columns 1-3
total.tab2[2,] <- c(mean(s1$units.total2[s1$pid7.sender==4 & s1$pid7>4], na.rm=T), # party of sender ==4 (ind), party of receiver >4 (rep)
                   mean(s1$units.total2[s1$pid7.sender==4 & s1$pid7==4], na.rm=T),# party of sender is == 4 (ind), party of receiver is 4 (Ind)
                   mean(s1$units.total2[s1$pid7.sender==4 & s1$pid7<4], na.rm=T)) # party of sender is ==4 (ind), party of receiver is <4 (dem)

### Row 3, columns 1-3
total.tab2[3,] <- c(mean(s1$units.total2[s1$pid7.sender<4 & s1$pid7>4], na.rm=T), # party of sender <4 (dem), party of receiver >4 (rep)
                   mean(s1$units.total2[s1$pid7.sender<4 & s1$pid7==4], na.rm=T),# party of sender is < 4 (dem), party of receiver is 4 (Ind)
                   mean(s1$units.total2[s1$pid7.sender<4 & s1$pid7<4], na.rm=T)) # party of sender is < 4 (dem), party of receiver is <4 (dem)

colnames(total.tab2) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(total.tab2) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=FALSE)
print(xtable(total.tab2, caption="Units of Information by Sender and Receiver Partisanship (Coder 2)", type="latex"), caption.placement="top")


```

\FloatBarrier

### Units of Information About the Economy

The coders also counted how many units of information in the messages were about the economy. The specific instructions were: "List all units of information contained in the message that are about the economy. This could include units about GDP, gross domestic product, unemployment, employment, inflation, economic policy, taxes, spending, manufacturing, jobs, economic growth, and any other topics related to the economy. Please copy the exact text used in the message or indicate which exact text suggested the unit. Leave blank if there are no units of information about the economy."

On average, messages contained `r round(mean(s1$units.econ, na.rm=T), digits=2)` units of information about the economy (Coder 1=`r round(mean(s1$units.econ1,  na.rm=T), digits=2)`, Coder 2=`r round(mean(s1$units.econ2,  na.rm=T), digits=2)`). 

### Units of Information Favoring and Against the Economy 

The coders also counted how many units of information favored the economy (described it as improving) or were against the economy (described it as declining). The specific instructions for favoring were: "This should be a subset of the units listed in Column F [units about the economy]. It should include the number of units of information about the economy that communicate a positive impression of the economy. This could include saying that the economy is growing, the economy is doing well, economic policies are working, GDP is growing, GDP is strong, the economy is strong, unemployment is down, etc." The instructions for coding units against the economy were: "This should be a subset of the units listed in Column F. It should include the number of units about the economy that communicate a negative impression of the economy. This could include saying that the economy is not growing, it's doing poorly, it's declining, shrinking, getting weaker, is weak, that economic policies are not working, GDP is not growing, GDP has not grown, GDP is weak, unemployment is up, etc." 

The table below shows the average number of units favoring and opposing the economy broken down by coder. While the coders reached slightly different numbers, the patterns are the same, such that there were more units favoring the economy than against the economy on average. 

```{r echo=FALSE, eval=TRUE, results="asis"}
# Favoring/Against Economy Table
econ.tab <- matrix(NA, nrow=3, ncol=2)
econ.tab[1,] <- c(mean(s1$favor.econ, na.rm=T), mean(s1$against.econ, na.rm=T))
econ.tab[2,] <- c(mean(s1$favor.econ1, na.rm=T), mean(s1$against.econ1, na.rm=T))
econ.tab[3,] <- c(mean(s1$favor.econ2, na.rm=T), mean(s1$against.econ2, na.rm=T))

colnames(econ.tab) <- c("Favoring", "Against")
rownames(econ.tab) <- c("Average", "Coder 1", "Coder 2")


options(xtable.comment=FALSE)
print(xtable(econ.tab, caption="Units of Information Favoring and Against the Economy by Coder", type="latex"), caption.placement="top")

```


### Units of Information with Factual Inaccuracies 

The coders were asked to fact-check the messages as well. The most common inaccuracies were individuals reporting that GDP declined when in fact it had increased or mis-reporting the GDP numbers as reported by the BEA. Factual inaccuracies were rare overall, but about 8 percent of the messages contained at least one inaccuracy (Coder 1=`r round(sum(prop.table(table(s1$wronginfo1))[2:4]), digits=3)*100` percent, Coder 2=`r round(sum(prop.table(table(s1$wronginfo2))[2:3]), digits=3)*100` percent). Given that the inaccuracies were few in number, I do not break this down by the partisanship of the senders and receivers. 

## Similarity Robustness Tests 

To demonstrate that the general patterns presented in the paper are not strictly driven by the similarity metric selected (cosine similarity), I present the similarity analyses using Jaccard Similarity and cosine similarity using doc 2 vec. I used the text2vec package in R (Selivanov \url{http://text2vec.org/}) and quanteda (Benoit et al. \url{https://quanteda.io/}). 

### Jaccard Similarity and Doc2Vec

```{r echo=FALSE, eval=TRUE}
# Similarity with text 2 vec SOCIAL MESSAGES
library(stringr)
library(text2vec)
prep_fun = function(x) {
  x %>% 
    # make text lower case
    str_to_lower %>% 
    # remove non-alphanumeric symbols
    str_replace_all("[^[:alnum:]]", " ") %>% 
    # collapse multiple spaces
    str_replace_all("\\s+", " ")
}
soc$message_clean = prep_fun(soc$message)

# Create two sets of documents on which we'll evaluate similarity
doc_set_1 = soc[-1,]
it1 = itoken(doc_set_1$message_clean, progressbar=FALSE)

doc_set_2 = soc[1,]
it2 = itoken(doc_set_2$message_clean, progressbar=FALSE)

# We need to define common space and project documents to it. This uses vocabulary-based vectorization for better interpretability 
it = itoken(soc$message_clean, progressbar=FALSE)
v = create_vocabulary(it) %>% prune_vocabulary(doc_proportion_max = 1, term_count_min = 5)
vectorizer = vocab_vectorizer(v)


# Jaccard similarity
  # For docs we measure it as proportion of number of common words to number of unique words in both documents. Particularly useful for duplicates detection. 
dtm1 = create_dtm(it1, vectorizer)
dtm2 = create_dtm(it2, vectorizer)
d1_d2_jac_sim = sim2(dtm1, dtm2, method="jaccard", norm="none")

doc_set_1$similarity.jaccard <- NA
for(i in 1:nrow(doc_set_1)){
  doc_set_1$similarity.jaccard[i] <- d1_d2_jac_sim[i,1]
}
#summary(doc_set_1$similarity.jaccard)

# Using cosine similarity but with text2vec
d1_d2_cos_sim = sim2(dtm1, dtm2, method="cosine", norm="l2")
doc_set_1$similarity.cosine <- NA
for(i in 1:nrow(doc_set_1)){
  doc_set_1$similarity.cosine[i] <- d1_d2_cos_sim[i,1]
}


```

```{r echo=FALSE, eval=TRUE}
# Similarity with text 2 vec NEWS ARTICLES

news$text_clean = prep_fun(news$text)

# Create two sets of documents on which we'll evaluate similarity
doc_set_1_news = news[2:nrow(news),]
it1 = itoken(doc_set_1_news$text_clean, progressbar=FALSE)

doc_set_2_news = news[1,]
it2 = itoken(doc_set_2_news$text_clean, progressbar=FALSE)

# We need to define common space and project documents to it. This uses vocabulary-based vectorization for better interpretability 
it = itoken(news$text_clean, progressbar=FALSE)
v = create_vocabulary(it) %>% prune_vocabulary(doc_proportion_max = 1, term_count_min = 5)
vectorizer = vocab_vectorizer(v)


# Jaccard similarity
  # For docs we measure it as proportion of number of common words to number of unique words in both documents. Particularly useful for duplicates detection. 
dtm1 = create_dtm(it1, vectorizer)
dtm2 = create_dtm(it2, vectorizer)
d1_d2_jac_sim_news = sim2(dtm1, dtm2, method="jaccard", norm="none")

doc_set_1_news$similarity.jaccard <- NA
for(i in 1:nrow(doc_set_1_news)){
  doc_set_1_news$similarity.jaccard[i] <- d1_d2_jac_sim_news[i,1]
}
#summary(doc_set_1_news$similarity.jaccard)

# Using cosine similarity but with text2vec
d1_d2_cos_sim_news = sim2(dtm1, dtm2, method="cosine", norm="l2")
doc_set_1_news$similarity.cosine <- NA
for(i in 1:nrow(doc_set_1_news)){
  doc_set_1_news$similarity.cosine[i] <- d1_d2_cos_sim_news[i,1]
}


```

```{r echo=FALSE, eval=TRUE, fig.height=4.5, fig.width=8, fig.cap="Distribution of similarity scores with cosine similarity and Jaccard similarity using doc 2 vec. While the shape of the curves is certainly different between Jaccard and cosine similarity, the main result is consistent: social messages are far less similar to news articles than news articles are to official reports."}
# Plot the distribution of similarity between news article and official report, then between social messages and news article
#par(mfrow=c(1,1))
par(mfrow=c(1,2))
plot(density(doc_set_1_news$similarity.cosine), main="Cosine Similarity (doc2vec)",
     lwd=3, lty=1, xlab="Cosine Similarity Score\n(1=Most Similar)",
     xlim=c(0,1), ylim=c(0,5))
lines(density(doc_set_1$similarity.cosine), lwd=3, lty=2)
legend(0, 5, legend=c("News vs. BEA", "Social Messages vs. Reuters"), lty=c(1,2), cex=.85, lwd=3, bty="n")

plot(density(doc_set_1_news$similarity.jaccard), main="Jaccard Similarity (doc2vec)",
     lwd=3, lty=1, xlab="Jaccard Similarity Score",
     ylim=c(0,13))
lines(density(doc_set_1$similarity.jaccard), lwd=3, lty=2)
legend(0.05, 12, legend=c("News vs. BEA", "Social Messages vs. Reuters"), lty=c(1,2), cex=.85, lwd=3, bty="n")
```

\FloatBarrier

### Manually Corrected Spelling 

```{r echo=FALSE, eval=TRUE}

soc$message.edited_clean = prep_fun(soc$message.edited)

# Create two sets of documents on which we'll evaluate similarity
doc_set_1_edited = soc[-1,]
it1 = itoken(doc_set_1_edited$message.edited_clean, progressbar=FALSE)

doc_set_2 = soc[1,] #same as before
it2 = itoken(doc_set_2$message_clean, progressbar=FALSE)

# We need to define common space and project documents to it. This uses vocabulary-based vectorization for better interpretability 
it = itoken(soc$message.edited_clean, progressbar=FALSE)
v = create_vocabulary(it) %>% prune_vocabulary(doc_proportion_max = 1, term_count_min = 5)
vectorizer = vocab_vectorizer(v)


# Jaccard similarity
  # For docs we measure it as proportion of number of common words to number of unique words in both documents. Particularly useful for duplicates detection. 
dtm1 = create_dtm(it1, vectorizer)
dtm2 = create_dtm(it2, vectorizer)
d1_d2_jac_sim_edited = sim2(dtm1, dtm2, method="jaccard", norm="none")

doc_set_1_edited$similarity.jaccard.edited <- NA
for(i in 1:nrow(doc_set_1_edited)){
  doc_set_1_edited$similarity.jaccard.edited[i] <- d1_d2_jac_sim_edited[i,1]
}
#summary(doc_set_1_news$similarity.jaccard)

# Using cosine similarity but with text2vec
d1_d2_cos_sim_edited = sim2(dtm1, dtm2, method="cosine", norm="l2")
doc_set_1_edited$similarity.cosine.edited <- NA
for(i in 1:nrow(doc_set_1_edited)){
  doc_set_1_edited$similarity.cosine.edited[i] <- d1_d2_cos_sim_edited[i,1]
}


```

```{r echo=FALSE, eval=TRUE, fig.height=4.5, fig.width=8, fig.cap="Distribution of similarity scores comparing social messages to the Reuters news article. The dashed line shows the scores using the unedited messages. The solid line shows the scores using messages that have been manually edited to correct spelling errors. The distributions are virtually identical, suggesting that typos and spelling errors were not biasing the similarity scores downward."}
# Plot the distribution of similarity scores with manually edited messages 

par(mfrow=c(1,2))
plot(density(doc_set_1_edited$similarity.cosine.edited), main="Cosine Similarity with Corrected Spelling\nSocial Messages vs. News Article", 
     xlab="Cosine Similarity Score", xlim=c(-.1,1),
     ylim=c(0,3),
     lwd=3, lty=1)
lines(density(doc_set_1$similarity.cosine), lwd=3, lty=3)
legend(-.1, 3.25, legend=c("Spelling Corrected", "Unedited"), lty=c(1,3), lwd=3, bty="n")

plot(density(doc_set_1_edited$similarity.jaccard.edited), main="Jaccard Similarity with Corrected Spelling\nSocial Messages vs. News Article", 
     xlab="Jaccard Similarity Score",
     lwd=3, lty=1)
lines(density(doc_set_1$similarity.jaccard), lwd=3, lty=3)
legend(.15, 7, legend=c("Spelling Corrected", "Unedited"), lty=c(1,3), lwd=3, bty="n", cex=.85)

```

\FloatBarrier

### Automated Summarizers 

I used two unique automated summarizers\footnote{I used \url{https://resoomer.com/en/} and \url{https://www.splitbrain.org/services/ots}.} to try to gauge whether the similarity results were simply a function of shorter documents as opposed to changing words. One summarizer, allowed me to indicate what percent of the original text length I wanted the summary to be. Given that the social messages in this experiment were very short, I used the automated summarizer to generate messages that were 10 percent of the original document length and 5 percent of the original document length. 

The table below shows the average cosine and Jaccard similarity scores for the social messages overall (.39 and .07 respectively), followed by the scores comparing the automated summarizers to the news article. I only generated one summary with each summarizer, so the scores in rows 2-4 compare one document to another, as opposed to reflecting the average over many comparisons.  The results suggest that the automated summaries are much more similar to the news article than the human generated social messages. 

The fifth row shows the scores when the social messages have been manually corrected for spelling, as illustrated in previous figures. These are almost identical to the original, unedited text, which means that the results are not driven by typos or spelling errors driving down the similarity scores. The 6th and 7th rows eliminate duplicates and re-calculate the similarity scores in both the original text (unedited) and edited for spelling corrections. 

```{r echo=FALSE, eval=TRUE}
# Similarity scores with automatic summarizers (first summarizer)

message.autosum1_clean = prep_fun(soc$message.autosum1)

media = prep_fun(soc$message[1])

# Create two sets of documents on which we'll evaluate similarity

it1 = itoken(message.autosum1_clean, progressbar=FALSE)

it2 = itoken(media, progressbar=FALSE)

# We need to define common space and project documents to it. This uses vocabulary-based vectorization for better interpretability 
it = itoken(c(message.autosum1_clean, media), progressbar=FALSE)
v = create_vocabulary(it) %>% prune_vocabulary(doc_proportion_max = 1, term_count_min = 5)
vectorizer = vocab_vectorizer(v)


# Jaccard similarity
  # For docs we measure it as proportion of number of common words to number of unique words in both documents. Particularly useful for duplicates detection. 
dtm1 = create_dtm(it1, vectorizer)
dtm2 = create_dtm(it2, vectorizer)
d1_d2_jac_sim_autosum1 = sim2(dtm1, dtm2, method="jaccard", norm="none")

# Using cosine similarity but with text2vec
d1_d2_cos_sim_autosum1 = sim2(dtm1, dtm2, method="cosine", norm="l2")


```

```{r echo=FALSE, eval=TRUE}
# Similarity scores with automatic summarizers (second summarizer, 5%)

message.autosum2.5pct_clean = prep_fun(soc$message.autosum2.5pct)

media = prep_fun(soc$message[1])

# Create two sets of documents on which we'll evaluate similarity

it1 = itoken(message.autosum2.5pct_clean, progressbar=FALSE)

it2 = itoken(media, progressbar=FALSE)

# We need to define common space and project documents to it. This uses vocabulary-based vectorization for better interpretability 
it = itoken(c(message.autosum2.5pct_clean, media), progressbar=FALSE)
v = create_vocabulary(it) %>% prune_vocabulary(doc_proportion_max = 1, term_count_min = 5)
vectorizer = vocab_vectorizer(v)


# Jaccard similarity
  # For docs we measure it as proportion of number of common words to number of unique words in both documents. Particularly useful for duplicates detection. 
dtm1 = create_dtm(it1, vectorizer)
dtm2 = create_dtm(it2, vectorizer)
d1_d2_jac_sim_autosum2.5pct = sim2(dtm1, dtm2, method="jaccard", norm="none")

# Using cosine similarity but with text2vec
d1_d2_cos_sim_autosum2.5pct = sim2(dtm1, dtm2, method="cosine", norm="l2")


```

```{r echo=FALSE, eval=TRUE}
# Similarity scores with automatic summarizers (second summarizer, 10%)

message.autosum2.10pct_clean = prep_fun(soc$message.autosum2.10pct)

media = prep_fun(soc$message[1])

# Create two sets of documents on which we'll evaluate similarity

it1 = itoken(message.autosum2.10pct_clean, progressbar=FALSE)

it2 = itoken(media, progressbar=FALSE)

# We need to define common space and project documents to it. This uses vocabulary-based vectorization for better interpretability 
it = itoken(c(message.autosum2.10pct_clean, media), progressbar=FALSE)
v = create_vocabulary(it) %>% prune_vocabulary(doc_proportion_max = 1, term_count_min = 5)
vectorizer = vocab_vectorizer(v)


# Jaccard similarity
  # For docs we measure it as proportion of number of common words to number of unique words in both documents. Particularly useful for duplicates detection. 
dtm1 = create_dtm(it1, vectorizer)
dtm2 = create_dtm(it2, vectorizer)
d1_d2_jac_sim_autosum2.10pct = sim2(dtm1, dtm2, method="jaccard", norm="none")

# Using cosine similarity but with text2vec
d1_d2_cos_sim_autosum2.10pct = sim2(dtm1, dtm2, method="cosine", norm="l2")



```


```{r echo=FALSE, eval=TRUE, results="asis"}
# Create a table showing the similarity scores with automatic summarizers 
sim.tab <- matrix(NA, nrow=7, ncol=2)
colnames(sim.tab) <- c("Cosine", "Jaccard")
rownames(sim.tab) <- c("Social Messages", "Automatic Summary 1", "Automatic Summary 2 (10pct article length)", "Automatic Summary 2 (5 pct article length)", "Social Messages Manually Corrected Spelling", "Unique Social Messages", "Unique Social Messages Manually Corrected Spelling")

sim.tab[1,] <- c(mean(doc_set_1$similarity.cosine, na.rm=T),
                 mean(doc_set_1$similarity.jaccard, na.rm=T))
sim.tab[2,] <- c(d1_d2_cos_sim_autosum1[1], d1_d2_jac_sim_autosum1[1])
sim.tab[3,] <- c(d1_d2_cos_sim_autosum2.10pct[1], d1_d2_jac_sim_autosum2.10pct[1])
sim.tab[4,] <- c(d1_d2_cos_sim_autosum2.5pct[1], d1_d2_jac_sim_autosum2.5pct[1])
sim.tab[5,] <- c(mean(doc_set_1_edited$similarity.cosine.edited, na.rm=T), mean(doc_set_1_edited$similarity.jaccard.edited, na.rm=T))
sim.tab[6,] <- c(mean(doc_set_1$similarity.cosine[!duplicated(doc_set_1$message)], na.rm=T), mean(doc_set_1$similarity.jaccard[!duplicated(doc_set_1$message)], na.rm=T))
sim.tab[7,] <- c(mean(doc_set_1_edited$similarity.cosine.edited[!duplicated(doc_set_1_edited$message)], na.rm=T), mean(doc_set_1_edited$similarity.jaccard.edited[!duplicated(doc_set_1_edited$message)], na.rm=T))

options(xtable.comment=FALSE)
print(xtable(sim.tab, caption="Similarity Score Robustness", type="latex"), caption.placement="top")

```

\FloatBarrier

### Similarity by Partisanship

```{r echo=FALSE, eval=TRUE, results="asis"}
# Mean similarity score by sender/receiver partisanship 
# Sender on rows (RID)
# Receiver on cols (RID)
# Use Jaccard similarity 

jacc.sim.party <- matrix(nrow=3, ncol=3)
# Republican senders
jacc.sim.party[1,] <- c(mean(doc_set_1$similarity.jaccard[doc_set_1$pid7>4 & doc_set_1$pid7.sender>4], na.rm=T),
                        mean(doc_set_1$similarity.jaccard[doc_set_1$pid7==4 & doc_set_1$pid7.sender>4], na.rm=T),
                        mean(doc_set_1$similarity.jaccard[doc_set_1$pid7<4 & doc_set_1$pid7.sender>4], na.rm=T))

# Independent senders
jacc.sim.party[2,] <- c(mean(doc_set_1$similarity.jaccard[doc_set_1$pid7>4 & doc_set_1$pid7.sender==4], na.rm=T),
                        mean(doc_set_1$similarity.jaccard[doc_set_1$pid7==4 & doc_set_1$pid7.sender==4], na.rm=T),
                        mean(doc_set_1$similarity.jaccard[doc_set_1$pid7<4 & doc_set_1$pid7.sender==4], na.rm=T))

# Democrat senders
jacc.sim.party[3,] <- c(mean(doc_set_1$similarity.jaccard[doc_set_1$pid7>4 & doc_set_1$pid7.sender<4], na.rm=T),
                        mean(doc_set_1$similarity.jaccard[doc_set_1$pid7==4 & doc_set_1$pid7.sender<4], na.rm=T),
                        mean(doc_set_1$similarity.jaccard[doc_set_1$pid7<4 & doc_set_1$pid7.sender<4], na.rm=T))

colnames(jacc.sim.party) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(jacc.sim.party) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=F)
print(xtable(jacc.sim.party, caption="Jaccard Similarity Between Social Messages and Reuters Article by Partisanship", type="latex"), caption.placement="top")

```

\FloatBarrier

## Sentiment Analysis 

In order to examine variation in the tone with which information was communicated, I estimate the sentiment of each document using a variety of sentiment dictionaries. It is unclear whether the sentiment would impact learning, but one could imagine that the sentiment could impact subjective evaluations. Specifically, more negative statements might lead one to have more negative evaluations of the economy and/or the President, compared to more positive evaluations. 


```{r echo=FALSE, eval=TRUE}

library(syuzhet)

# Hu & Liu
soc$sentiment.bing <- NA
soc$sentiment.bing <- get_sentiment(as.character(soc$message), method="bing")

# AFinn
soc$sentiment.afinn <- NA
soc$sentiment.afinn <- get_sentiment(as.character(soc$message), method="afinn")

# Mohammad
soc$sentiment.nrc <- NA
soc$sentiment.nrc <- get_sentiment(as.character(soc$message), method="nrc")

```

```{r sentiment_author, echo=FALSE, eval=TRUE, fig.height=10, fig.width=6.5, fig.cap="Sentiment Scores by Dictionary and Partisanship of Informant. The vertical dashed line indicates the sentiment score for the media article."}


# Plot the distribution of similarity scores by dictionary

par(mfrow=c(3,1))
plot(density(s1$sentiment.bing[s1$treatment=="socialind"], na.rm=T),
     main="Hu and Liu (2004) Sentiment Scores\nBy Treatment Condition",
     xlab="Sentiment Score", ylab="Density",
     lwd=2, col="seashell4",
     xlim=c(-15,15), ylim=c(0,.35))
lines(density(s1$sentiment.bing[s1$treatment=="socialrep"], na.rm=T), lwd=2, col="firebrick")
lines(density(s1$sentiment.bing[s1$treatment=="socialdem"], na.rm=T), lwd=2, col="dodgerblue2")
abline(v=mean(soc$sentiment.bing[1], na.rm=T),
       lwd=2, lty=2)
legend(5, .3, legend=c("Republican Informant", "Democrat Informant", "Independent Informant"), fill=c("firebrick", "dodgerblue2", "seashell4"), bty="n", cex=.8)

plot(density(s1$sentiment.afinn[s1$treatment=="socialrep"], na.rm=T),
     main="Finn (2011) Sentiment Scores\nBy Treatment Condition",
     xlab="Sentiment Score", ylab="Density",
     lwd=2, col="firebrick",
     xlim=c(-15,15), ylim=c(0,.35))
lines(density(s1$sentiment.afinn[s1$treatment=="socialind"], na.rm=T), lwd=2, col="seashell4")
lines(density(s1$sentiment.afinn[s1$treatment=="socialdem"], na.rm=T), lwd=2, col="dodgerblue2")
abline(v=mean(soc$sentiment.afinn[1], na.rm=T),
       lwd=2, lty=2)
legend(5, .3, legend=c("Republican Informant", "Democrat Informant", "Independent Informant"), fill=c("firebrick", "dodgerblue2", "seashell4"), bty="n", cex=.8)

plot(density(s1$sentiment.nrc[s1$treatment=="socialrep"], na.rm=T),
     main="Mohammad (2010) Sentiment Scores\nBy Treatment Condition",
     xlab="Sentiment Score", ylab="Density",
     lwd=2, col="firebrick",
     xlim=c(-15,15), ylim=c(0,.35))
lines(density(s1$sentiment.nrc[s1$treatment=="socialind"], na.rm=T), lwd=2, col="seashell4")
lines(density(s1$sentiment.nrc[s1$treatment=="socialdem"], na.rm=T), lwd=2, col="dodgerblue2")
abline(v=mean(soc$sentiment.nrc[1], na.rm=T),
       lwd=2, lty=2)
legend(5, .3, legend=c("Republican Informant", "Democrat Informant", "Independent Informant"), fill=c("firebrick", "dodgerblue2", "seashell4"), bty="n", cex=.8)

```

The figure below shows the distribution of sentiment scores based on the partisanship of the author of the message across three commonly used sentiment dictionaries. The red lines represent messages written by Republicans, the blue lines represent messages written by Democrats, and the gray lines represent messages written by Independents. The vertical dashed line in each figure is the sentiment score assigned to the media article. The top figure, using the Hu and Liu (2004) sentiment dictionary, is the only one for which the media article was significantly more negative than the social messages. 

Using the Finn (2011) and Mohammad (2010) scores, however, we see a somewhat different pattern with respect to the social messages compared to the media messages. Within the social messages, the comparisons were similar across all three dictionaries, though. 

```{r sentiment_recipient, echo=FALSE, eval=TRUE, fig.height=10, fig.width=6.5, fig.cap="Distribution of sentiment scores by informant and recipient partisanship across sentiment dictionaries."}

# Plot the distribution of similarity scores by dictionary

par(mfrow=c(3,1))
plot(density(s1$sentiment.bing[s1$pid7==4], na.rm=T),
     main="Hu and Liu (2004) Sentiment Scores\nBy Recipient Partisanship",
     xlab="Sentiment Score", ylab="Density",
     lwd=2, col="seashell4",
     xlim=c(-15,15), ylim=c(0,.5))
lines(density(s1$sentiment.bing[s1$pid7>4], na.rm=T), lwd=2, col="firebrick")
lines(density(s1$sentiment.bing[s1$pid7<4], na.rm=T), lwd=2, col="dodgerblue2")
abline(v=mean(soc$sentiment.bing[1], na.rm=T),
       lwd=2, lty=2)
legend(5, .3, legend=c("Republican Informant", "Democrat Informant", "Independent Informant"), fill=c("firebrick", "dodgerblue2", "seashell4"), bty="n", cex=.8)

plot(density(s1$sentiment.afinn[s1$pid7>4], na.rm=T),
     main="Finn (2011) Sentiment Scores\nBy Recipient Partisanship",
     xlab="Sentiment Score", ylab="Density",
     lwd=2, col="firebrick",
     xlim=c(-15,15), ylim=c(0,.5))
lines(density(s1$sentiment.afinn[s1$pid7==4], na.rm=T), lwd=2, col="seashell4")
lines(density(s1$sentiment.afinn[s1$pid7<4], na.rm=T), lwd=2, col="dodgerblue2")
abline(v=mean(soc$sentiment.afinn[1], na.rm=T),
       lwd=2, lty=2)
legend(5, .3, legend=c("Republican Informant", "Democrat Informant", "Independent Informant"), fill=c("firebrick", "dodgerblue2", "seashell4"), bty="n", cex=.8)

plot(density(s1$sentiment.nrc[s1$pid7>4], na.rm=T),
     main="Mohammad (2010) Sentiment Scores\nBy Recipient Partisanship",
     xlab="Sentiment Score", ylab="Density",
     lwd=2, col="firebrick",
     xlim=c(-15,15), ylim=c(0,.5))
lines(density(s1$sentiment.nrc[s1$pid7==4], na.rm=T), lwd=2, col="seashell4")
lines(density(s1$sentiment.nrc[s1$pid7<4], na.rm=T), lwd=2, col="dodgerblue2")
abline(v=mean(soc$sentiment.nrc[1], na.rm=T),
       lwd=2, lty=2)
legend(5, .3, legend=c("Republican Recipient", "Democrat Recipient", "Independent Recipient"), fill=c("firebrick", "dodgerblue2", "seashell4"), bty="n", cex=.8)

```

\FloatBarrier

The following tables show the sentiment scores from each dictionary broken down by the partisanship of the informant and the recipient. The only consistent pattern is that Republicans sent the most positive messages, regardless of the partisanship of the recipient. The only exception to this is with the Mohammad (2010) sentiment scores where Independents sent more positive messages to Republicans than Republicans sent to other Republicans.

```{r echo=FALSE, eval=TRUE, results="asis"}

huliu_party <- matrix(NA, nrow=3, ncol=3)
#Republican Senders
huliu_party[1,] <- c(mean(s1$sentiment.bing[s1$pid7.sender>4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.bing[s1$pid7.sender>4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.bing[s1$pid7.sender>4 & s1$pid7<4], na.rm=T))

#Independent Senders
huliu_party[2,] <- c(mean(s1$sentiment.bing[s1$pid7.sender==4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.bing[s1$pid7.sender==4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.bing[s1$pid7.sender==4 & s1$pid7<4], na.rm=T))

#Democrat Senders
huliu_party[3,] <- c(mean(s1$sentiment.bing[s1$pid7.sender<4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.bing[s1$pid7.sender<4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.bing[s1$pid7.sender<4 & s1$pid7<4], na.rm=T))

colnames(huliu_party) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(huliu_party) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=F)
print(xtable(huliu_party, caption="Sentiment Scores by Party - Hu and Liu (2004)", type="latex"),
      caption.placement="top")
```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, results="asis"}

afinn_party <- matrix(NA, nrow=3, ncol=3)
#Republican Senders
afinn_party[1,] <- c(mean(s1$sentiment.afinn[s1$pid7.sender>4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.afinn[s1$pid7.sender>4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.afinn[s1$pid7.sender>4 & s1$pid7<4], na.rm=T))

#Independent Senders
afinn_party[2,] <- c(mean(s1$sentiment.afinn[s1$pid7.sender==4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.afinn[s1$pid7.sender==4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.afinn[s1$pid7.sender==4 & s1$pid7<4], na.rm=T))

#Democrat Senders
afinn_party[3,] <- c(mean(s1$sentiment.afinn[s1$pid7.sender<4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.afinn[s1$pid7.sender<4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.afinn[s1$pid7.sender<4 & s1$pid7<4], na.rm=T))

colnames(afinn_party) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(afinn_party) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=F)
print(xtable(afinn_party, caption="Sentiment Scores by Party - Finn (2011)", type="latex"),
      caption.placement="top")
```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, results="asis"}

nrc_party <- matrix(NA, nrow=3, ncol=3)
#Republican Senders
nrc_party[1,] <- c(mean(s1$sentiment.nrc[s1$pid7.sender>4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.nrc[s1$pid7.sender>4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.nrc[s1$pid7.sender>4 & s1$pid7<4], na.rm=T))

#Independent Senders
nrc_party[2,] <- c(mean(s1$sentiment.nrc[s1$pid7.sender==4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.nrc[s1$pid7.sender==4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.nrc[s1$pid7.sender==4 & s1$pid7<4], na.rm=T))

#Democrat Senders
nrc_party[3,] <- c(mean(s1$sentiment.nrc[s1$pid7.sender<4 & s1$pid7>4], na.rm=T),
                     mean(s1$sentiment.nrc[s1$pid7.sender<4 & s1$pid7==4], na.rm=T),
                     mean(s1$sentiment.nrc[s1$pid7.sender<4 & s1$pid7<4], na.rm=T))

colnames(nrc_party) <- c("Rep Receiver", "Ind Receiver", "Dem Receiver")
rownames(nrc_party) <- c("Rep Sender", "Ind Sender", "Dem Sender")

options(xtable.comment=F)
print(xtable(nrc_party, caption="Sentiment Scores by Party - Mohammad (2010)", type="latex"),
      caption.placement="top")
```

\FloatBarrier

## Individual Differences Explaining Text Characteristics

The following tables examine the relationship between pre-treatment individual characteristics and the text generated by Study 1 respondents as they wrote their messages. 

```{r echo=FALSE, eval=TRUE}

# Word Count
mod.wordcount <- lm(wordcount ~ copartisan.sender + interest.sender_r + female.sender + ideology.sender, data=s1)

# Similarity
mod.similarity <- lm(similarity ~ copartisan.sender + interest.sender_r + female.sender + ideology.sender, data=s1)
mod.cosine <- lm(similarity.cosine ~ copartisan.sender + interest.sender_r + female.sender + ideology.sender, data=doc_set_1[!duplicated(doc_set_1$message),])
mod.jaccard <- lm(similarity.jaccard ~ copartisan.sender + interest.sender_r + female.sender + ideology.sender, data=doc_set_1[!duplicated(doc_set_1$message),])

# Sentiment
mod.bing <- lm(sentiment.bing ~ copartisan.sender + interest.sender_r + female.sender + ideology.sender, data=s1)
mod.afinn <- lm(sentiment.afinn ~ copartisan.sender +  interest.sender_r + female.sender + ideology.sender, data=s1)
mod.nrc <- lm(sentiment.nrc ~ copartisan.sender + interest.sender_r + female.sender + ideology.sender, data=s1)
```

```{r echo=FALSE, eval=TRUE, results="asis"}

stargazer(mod.wordcount,
          header=FALSE, font.size="footnotesize",
          covariate.labels=c("Copartisan with Recipient", "Political Interest", "Female", "Ideology (conservative)"),
          title="Relationship Between Informant Characteristics and Length of Messages Transmitted",
          dep.var.labels=c("Word Count"))
```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, results="asis"}

stargazer(mod.similarity, mod.cosine, mod.jaccard,
          header=FALSE, font.size="footnotesize",
          covariate.labels=c("Copartisan with Recipient", "Political Interest", "Female", "Ideology (conservative)"),
          title="Relationship Between Informant Characteristics and Similarity of Messages Transmitted",
          dep.var.labels=c("Cosine", "Cosine (doc2vec)", "Jaccard (doc2vec)"))
```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, results="asis"}

stargazer(mod.bing, mod.afinn, mod.nrc,
          header=FALSE, font.size="footnotesize",
          covariate.labels=c("Copartisan with Recipient", "Political Interest", "Female", "Ideology (conservative)"),
          title="Relationship Between Informant Characteristics and Sentiment of Messages Transmitted",
          dep.var.labels=c("Hu Liu", "Finn", "Mohammad"))
```

\FloatBarrier

The following tables examine the relationship between individual characteristics and the text they generated in writing their messages. In these tables, some variables are \textit{post-treatment}, meaning that the results should be interpreted with caution (Montgomery, Nyhan, and Torres 2018). 

```{r echo=FALSE, eval=TRUE, results="asis"}

# Some post-treatment covariates included

s2 <- read.csv("grapevine_study2.csv")

# Word Count
mod.wordcount.unique <- lm(wordcount ~ copartisan.sender + correct.sender + interest.sender_r + often.ask.sender_r + female.sender + bias.sender + trust.sender_r + inform.sender, data=s2)

# Similarity
mod.similarity.unique <- lm(similarity ~ copartisan.sender + correct.sender + interest.sender_r + often.ask.sender_r + female.sender + bias.sender + trust.sender_r + inform.sender, data=s2)

# Sentiment
mod.bing.unique <- lm(sentiment.bing ~ copartisan.sender + correct.sender + interest.sender_r + often.ask.sender_r + female.sender + bias.sender + trust.sender_r + inform.sender, data=s2)
mod.afinn.unique <- lm(sentiment.afinn ~ copartisan.sender + correct.sender + interest.sender_r + often.ask.sender_r + female.sender + bias.sender + trust.sender_r + inform.sender, data=s2)
mod.nrc.unique <- lm(sentiment.nrc ~ copartisan.sender + correct.sender + interest.sender_r + often.ask.sender_r + female.sender + bias.sender + trust.sender_r + inform.sender, data=s2)

stargazer(mod.wordcount.unique,
          header=FALSE, font.size="footnotesize",
          covariate.labels=c("Copartisan with Recipient", "Knowledge from Article", "Political Interest", "How Often Asked for Info", "Female", "Perceived Bias in Article (lib-cons)", "Trust in Accuracy of Article", "Tried to Objectively Inform Recipient"),
          title="Relationship Between Informant Characteristics and Length of Messages Transmitted",
          dep.var.labels=c("Word Count"))

stargazer(mod.similarity.unique,
          header=FALSE, font.size="footnotesize",
          covariate.labels=c("Copartisan with Recipient", "Knowledge from Article", "Political Interest", "How Often Asked for Info", "Female", "Perceived Bias in Article (lib-cons)", "Trust in Accuracy of Article", "Tried to Objectively Inform Recipient"),
          title="Relationship Between Informant Characteristics and Similarity of Messages Transmitted",
          dep.var.labels=c("Cosine", "Cosine (doc2vec)", "Jaccard (doc2vec)"))

stargazer(mod.bing.unique, mod.afinn.unique, mod.nrc.unique,
          header=FALSE, font.size="footnotesize",
          covariate.labels=c("Copartisan with Recipient", "Knowledge from Article", "Political Interest", "How Often Asked for Info", "Female", "Perceived Bias in Article (lib-cons)", "Trust in Accuracy of Article", "Tried to Objectively Inform Recipient"),
          title="Relationship Between Informant Characteristics and Sentiment of Messages Transmitted",
          dep.var.labels=c("Hu Liu", "Finn", "Mohammad"))
```

\FloatBarrier

\clearpage

# Study 2

## Sample Characteristics

```{r study2sample, echo=FALSE, eval=TRUE, results="asis"}
s2$rep <- NA
s2$rep[s2$pid7>4] <- 1
s2$rep[s2$pid7<=4] <- 0

s2$ind <- NA
s2$ind[s2$pid7==4] <- 1
s2$ind[s2$pid7!=4] <- 0

s2$college <- NA
s2$college[s2$edu>=4] <- 1
s2$college[s2$edu<4] <- 0

s2$income100k <- NA
s2$income100k[s2$income>5] <- 1
s2$income100k[s2$income<=5] <- 0

# Demographics to include 
s2demog <- c("dem", "rep", "ind", "female", "white", "age", "ideology", "college", "income100k")

study2sample <- matrix(NA, nrow=length(s2demog), ncol=1)
for(i in 1:nrow(study2sample)){
  study2sample[i,] <- mean(s2[, s2demog[i]], na.rm=T)
}

rownames(study2sample) <- c("Democrat", "Republican", "Independent", "Female", "White", "Age", "Ideology", "College Degree", "Income over 100k")
colnames(study2sample) <- "Prop/Mean"

options(xtable.comment=FALSE)
print(xtable(study2sample, caption="Sample Characteristics for Study 2", type="latex"),
      caption.placement="top")

```

\FloatBarrier

## Balance Tables

```{r study2balance, echo=FALSE, eval=TRUE, results="asis"}
study2balance <- matrix(NA, nrow=length(s2demog), ncol=4)
for(i in 1:nrow(study2balance)){
  study2balance[i,] <- c(mean(s2[, s2demog[i]][s2$treatment=="media"], na.rm=T),
                         mean(s2[, s2demog[i]][s2$treatment=="socialdem"], na.rm=T),
                         mean(s2[, s2demog[i]][s2$treatment=="socialind"], na.rm=T),
                         mean(s2[, s2demog[i]][s2$treatment=="socialrep"], na.rm=T))
}

rownames(study2balance) <- c("Democrat", "Republican", "Independent", "Female", "White", "Age", "Ideology", "College Degree", "Income Over 100k")
colnames(study2balance) <- c("Media", "Democrat Informant", "Independent Informant", "Republican Informant")

options(xtable.comment=FALSE)
print(xtable(study2balance, caption="Balance Between Treatment Groups in Study 2", type="latex"),
      caption.placement="top")
```

\FloatBarrier

```{r study2balance_ideal, echo=FALSE, eval=TRUE, results="asis"}

s2$ideal.informant <- NA
s2$ideal.informant[s2$copartisan.sender==1 & s2$knowledgeable.sender==1] <- 1
s2$ideal.informant[s2$copartisan.sender==0 & s2$knowledgeable.sender==0] <- 0
s2$ideal.informant[s2$copartisan.sender==1 & s2$knowledgeable.sender==0] <- 0
s2$ideal.informant[s2$copartisan.sender==0 & s2$knowledgeable.sender==1] <- 0

study2balance_ideal <- matrix(NA, nrow=length(s2demog), ncol=3)
for(i in 1:nrow(study2balance_ideal)){
  study2balance_ideal[i,] <- c(mean(s2[, s2demog[i]][s2$ideal.informant==1], na.rm=T),
                         mean(s2[, s2demog[i]][s2$ideal.informant==0], na.rm=T),
                         mean(s2[, s2demog[i]][s2$treatment=="media"], na.rm=T))
}

rownames(study2balance_ideal) <- c("Democrat", "Republican", "Independent", "Female", "White", "Age", "Ideology", "College Degree", "Income Over 100k")
colnames(study2balance_ideal) <- c("Ideal Informant", "Non-Ideal Informant", "Media")

options(xtable.comment=FALSE)
print(xtable(study2balance_ideal, caption="Balance Between Ideal Informants and Non-Ideal Informants in Study 2", type="latex"),
      caption.placement="top")
```

\FloatBarrier

## Research Design Illustration 

\begin{figure}[h]
\centering
\includegraphics[width=7.5in]{figures/researchdesign.pdf}
\caption{Illustration of Study 2 research design. Dashed lines represent random assignment and solid lines represent non-random assignment. The illustration reflects the four stages of the study: pre-treatment questions, distractor questions, information treatments, and post-treatment questions.}
\label{researchdesign}
\end{figure}

\FloatBarrier

## Learning by Informant Type 

```{r echo=FALSE, eval=TRUE, fig.cap="Change in number of correct responses by informant type."}
par(mfrow=c(1,1))
labs <- c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant")

labs2 <- c("Copartisan", "Outpartisan", "Knows More", "Knows Less", "Ideal", "Non-Ideal")
infochange.means2 <- c(mean(s2$correct.change[s2$copartisan.sender==1], na.rm=T),
                      mean(s2$correct.change[s2$copartisan.sender==0], na.rm=T),
                      mean(s2$correct.change[s2$knowledgeable.sender==1], na.rm=T),
                mean(s2$correct.change[s2$knowledgeable.sender==0], na.rm=T), 
                mean(s2$correct.change[s2$ideal.informant==1], na.rm=T),
                mean(s2$correct.change[s2$ideal.informant==0],na.rm=T))

infochange.mat2 <- matrix(NA, nrow=6, ncol=2)
infochange.mat2[,1] <- c(1,2,3,4,5,6)
infochange.mat2[,2] <- infochange.means2

infochange.se2 <- c(sd(s2$correct.change[s2$copartisan.sender==1], na.rm=T)/sqrt(table(s2$copartisan.sender)[2]),
                   sd(s2$correct.change[s2$copartisan.sender==0], na.rm=T)/sqrt(table(s2$copartisan.sender)[1]),
                   sd(s2$correct.change[s2$knowledgeable.sender==1], na.rm=T)/sqrt(table(s2$knowledgeable.sender)[2]),
                   sd(s2$correct.change[s2$knowledgeable.sender==0], na.rm=T)/sqrt(table(s2$knowledgeable.sender)[1]),
                   sd(s2$correct.change[s2$ideal.informant==1], na.rm=T)/sqrt(table(s2$ideal.informant)[2]),
                   sd(s2$correct.change[s2$ideal.informant==0], na.rm=T)/sqrt(table(s2$ideal.informant)[1]))

x <- 1:6
par(mar=c(5,6,4,2)+.1)
plot(infochange.means2, x, main="Information Recall by Informant Type",
     xlim=c(-1,1), xlab="Change in Recall (Post - Pre)", ylab="",
     axes=FALSE,
     col=c("black", "black", "black", "black", "black", "black"),
     pch=19)
axis(1)
axis(2, at=1:length(labs), las=1, labels=labs2)
arrows(infochange.means2-1.96*infochange.se2, x, infochange.means2+1.96*infochange.se2, x, angle=90, code=.5, col=c("black", "black", "black", "black", "black", "black"))
abline(v=0)
text(.55,6.10,"More Correct Recall Post-Treatment", cex=.75)
text(-.55,6.10,"Less Correct Recall Post-Treatment", cex=.75)
box()


```

\FloatBarrier

## Learning with Controls

```{r learning_ideal, echo=FALSE, eval=TRUE, results="asis"}

s2$media <- NA
s2$media[s2$treatment=="media"] <- 1
s2$media[s2$treatment!="media"] <- 0
#table(s2$media)

s2$goodinf <- NA
s2$goodinf[s2$media==1] <- 0
s2$goodinf[s2$ideal.informant==1] <- 1
s2$goodinf[s2$ideal.informant==0] <- 0
#table(s2$goodinf)

s2$badinf <- NA
s2$badinf[s2$media==1] <- 0
s2$badinf[s2$ideal.informant==1] <- 0
s2$badinf[s2$ideal.informant==0] <- 1
#table(s2$badinf)

base.inf <- lm(correct.change ~ goodinf + badinf, data=s2)
base.controls.inf <- lm(correct.change ~ goodinf + badinf + dem + age + white + female + edu + polknow + interest, data=s2) 
soc.base <- lm(correct.change ~ ideal.informant, data=s2)
soc.base.textcontrols <- lm(correct.change ~ ideal.informant + wordcount + similarity + pct.favor.econ, data=s2)
soc.base.recipcontrols <- lm(correct.change ~ ideal.informant + wordcount + similarity + pct.favor.econ + dem + age + white + female + edu + polknow + interest, data=s2)

stargazer(base.inf, base.controls.inf,
          header=FALSE, font.size="footnotesize",
          title="Change in Information Recall: Informant Type vs. Media",
          dep.var.labels="Change in Number of Correct Answers (Post-Pre)",
          covariate.labels=c("Ideal Informant", "Non-Ideal Informant", "Democrat", "Age", "White", "Female", "Education", "Political Knowledge", "Political Interest"))

```


\FloatBarrier

```{r learning_v2, echo=FALSE, eval=TRUE, results="asis"}

stargazer(soc.base, soc.base.textcontrols, soc.base.recipcontrols,
          header=FALSE, font.size="footnotesize",
          title="Change in Information Recall: Ideal vs. Non-Ideal Informant",
          dep.var.labels="Change in Number of Correct Answers (Post-Pre)",
          covariate.labels=c("Ideal Informant", "Word Count", "Similarity", "Sentiment (Hand-coded)", "Democrat", "Age", "White", "Female", "Education", "Political Knowledge", "Political Interest"))

```

\FloatBarrier


## Ceiling Effects

This section examines the possibility of ceiling effects driving the results presented in the paper. For example, when examining learning, are the results simply driven by respondents doing very well pre-treatment thus having little room to do better after exposure to information? When examining change in presidential approval, is approval so high that the null results are driven by respondents having nowhere higher to move on the approval scale? The following figures present the treatment effects presented in the paper examining subsets of respondents based on their \textit{pre-treatment} measures. 

### Learning 

Overall, looking at learning, we see that those who answered fewer questions correctly pre-treatment were did indeed learn in all treatment conditions [regardless of their randomly assigned informational message]. The table below shows the proportion of respondents who answered each question correctly pre-treatment and post-treatment. 

```{r propcorrect, echo=FALSE, eval=TRUE, results="asis"}

questioncorrect <- matrix(NA, nrow=6, ncol=2)
questioncorrect[1,1:2] <- c(prop.table(table(s2$econgrew.post.correct[s2$econgrew.pre.correct==0]))["1"],
prop.table(table(s2$econgrew.post.correct[s2$econgrew.pre.correct==1]))["1"])
questioncorrect[2,1:2] <- c(prop.table(table(s2$gdpgrew.post.correct[s2$gdpgrew.pre.correct==0]))["1"],
prop.table(table(s2$gdpgrew.post.correct[s2$gdpgrew.pre.correct==1]))["1"])
questioncorrect[3,1:2] <- c(prop.table(table(s2$gdpjanmarch.post.correct[s2$gdpjanmarch.pre.correct==0]))["1"],
prop.table(table(s2$gdpjanmarch.post.correct[s2$gdpjanmarch.pre.correct==1]))["1"])
questioncorrect[4,1:2] <- c(prop.table(table(s2$trumptarget.post.correct[s2$trumptarget.pre.correct==0]))["1"],
prop.table(table(s2$trumptarget.post.correct[s2$trumptarget.pre.correct==1]))["1"])
questioncorrect[5,1:2] <- c(prop.table(table(s2$avgrate.post.correct[s2$avgrate.pre.correct==0]))["1"],
prop.table(table(s2$avgrate.post.correct[s2$avgrate.pre.correct==1]))["1"])
questioncorrect[6,1:2] <- 
c(prop.table(table(s2$conspending.post.correct[s2$conspending.pre.correct==0]))["1"],
prop.table(table(s2$conspending.post.correct[s2$conspending.pre.correct==1]))["1"])

rownames(questioncorrect) <- c("Economy Grew", "GDP Grew", "GDP Jan-March", "Trump Target", "Average Rate", "Consumer Spending")
colnames(questioncorrect) <- c("Wrong Pre, Right Post", "Right Pre, Right Post")
options(xtable.comment=FALSE)
print(xtable(questioncorrect, caption="Proportion of Respondents Answering Questions Correctly Post-Treatment based on Pre-Treatment Performance", type="latex"), caption.placement="top", size="footnotesize")
```

\FloatBarrier

The figures that follow show the treatment effects broken down by subsets of respondents based on pre-treatment responses. Among respondents who answered no better than chance pre-treatment (0/6-3/6), receiving information from the media led to significantly more learning than receiving information from an ideal informant. Curiously, receiving information from non-ideal informants was statistically indistinguishable from receiving information from the media. 

```{r echo=FALSE, eval=TRUE, fig.cap="Change in number of correct responses among those who answered fewer than three questions correctly pre-treatment."}

s2sub <- subset(s2, s2$correct.pre<3)
infochange.means <- c(mean(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T),
                      mean(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T),
                      mean(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T))
infochange.mat <- matrix(NA, nrow=6, ncol=2)
infochange.mat[,1] <- c(1,2,3,4,5,6)
infochange.mat[,2] <- infochange.means

infochange.se <- c(sd(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T)/sqrt(table(s2sub$ideal.informant)[1]),
                   sd(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T)/sqrt(table(s2sub$ideal.informant)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T)/sqrt(table(s2sub$treatment)[1]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(table(s2sub$treatment)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T)/sqrt(table(s2sub$treatment)[3]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(table(s2sub$treatment)[4]))

labs <- c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant")
x <- 1:6
par(mar=c(5,6,4,2)+.1)
plot(infochange.means, x, main="Information Recall by Treatment Group\nThose Who Got 0/6-2/6 Pre Treatment",
     xlim=c(-1,2), xlab="Change in Recall (Post - Pre)", ylab="",
     axes=FALSE,
     col=c("black", "black", "seagreen", "black", "black", "black"),
     pch=19)
axis(1)
axis(2, at=1:length(labs), las=1, labels=labs)
arrows(infochange.means-1.96*infochange.se, x, infochange.means+1.96*infochange.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
#text(.55,6.10,"More Correct Recall Post-Treatment", cex=.75)
#text(-.55,6.10,"Less Correct Recall Post-Treatment", cex=.75)
box()


```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, fig.cap="Change in number of correct responses among those who answered fewer than four questions correctly pre-treatment."}
s2sub <- subset(s2, s2$correct.pre<4)
infochange.means <- c(mean(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T),
                      mean(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T),
                      mean(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T))
infochange.mat <- matrix(NA, nrow=6, ncol=2)
infochange.mat[,1] <- c(1,2,3,4,5,6)
infochange.mat[,2] <- infochange.means

infochange.se <- c(sd(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T)/sqrt(table(s2sub$ideal.informant)[1]),
                   sd(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T)/sqrt(table(s2sub$ideal.informant)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T)/sqrt(table(s2sub$treatment)[1]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(table(s2sub$treatment)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T)/sqrt(table(s2sub$treatment)[3]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(table(s2sub$treatment)[4]))

labs <- c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant")
x <- 1:6
par(mar=c(5,6,4,2)+.1)
plot(infochange.means, x, main="Information Recall by Treatment Group\nThose Who Got 0/6-3/6 Pre Treatment",
     xlim=c(-1,2), xlab="Change in Recall (Post - Pre)", ylab="",
     axes=FALSE,
     col=c("black", "black", "seagreen", "black", "black", "black"),
     pch=19)
axis(1)
axis(2, at=1:length(labs), las=1, labels=labs)
arrows(infochange.means-1.96*infochange.se, x, infochange.means+1.96*infochange.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
#text(.55,6.10,"More Correct Recall Post-Treatment", cex=.75)
#text(-.55,6.10,"Less Correct Recall Post-Treatment", cex=.75)
box()


```

\FloatBarrier

```{r echo=FALSE, eval=TRUE, fig.cap="Change in number of correct responses among those who answered fewer than 5 questions correctly pre-treatment."}
s2sub <- subset(s2, s2$correct.pre<5)
infochange.means <- c(mean(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T),
                      mean(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T),
                      mean(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T))
infochange.mat <- matrix(NA, nrow=6, ncol=2)
infochange.mat[,1] <- c(1,2,3,4,5,6)
infochange.mat[,2] <- infochange.means

infochange.se <- c(sd(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T)/sqrt(table(s2sub$ideal.informant)[1]),
                   sd(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T)/sqrt(table(s2sub$ideal.informant)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T)/sqrt(table(s2sub$treatment)[1]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(table(s2sub$treatment)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T)/sqrt(table(s2sub$treatment)[3]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(table(s2sub$treatment)[4]))


labs <- c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant")
x <- 1:6
par(mar=c(5,6,4,2)+.1)
plot(infochange.means, x, main="Information Recall by Treatment Group\nThose Who Got 0/6-4/6 Pre Treatment",
     xlim=c(-1,2), xlab="Change in Recall (Post - Pre)", ylab="",
     axes=FALSE,
     col=c("black", "black", "seagreen", "black", "black", "black"),
     pch=19)
axis(1)
axis(2, at=1:length(labs), las=1, labels=labs)
arrows(infochange.means-1.96*infochange.se, x, infochange.means+1.96*infochange.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
#text(.55,6.10,"More Correct Recall Post-Treatment", cex=.75)
#text(-.55,6.10,"Less Correct Recall Post-Treatment", cex=.75)
box()


```

\FloatBarrier 

```{r echo=FALSE, eval=TRUE, fig.cap="Change in number of correct responses among those who answered at least one question incorrectly pre-treatment."}
s2sub <- subset(s2, s2$correct.pre<6)
infochange.means <- c(mean(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T),
                      mean(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T),
                      mean(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T))
infochange.mat <- matrix(NA, nrow=6, ncol=2)
infochange.mat[,1] <- c(1,2,3,4,5,6)
infochange.mat[,2] <- infochange.means

infochange.se <- c(sd(s2sub$correct.change[s2sub$ideal.informant==0], na.rm=T)/sqrt(table(s2sub$ideal.informant==0)[1]),
                   sd(s2sub$correct.change[s2sub$ideal.informant==1], na.rm=T)/sqrt(table(s2sub$ideal.informant)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="media"], na.rm=T)/sqrt(table(s2sub$treatment)[1]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(table(s2sub$treatment)[2]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialind"], na.rm=T)/sqrt(table(s2sub$treatment)[3]),
                   sd(s2sub$correct.change[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(table(s2sub$treatment)[4]))


labs <- c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant")
x <- 1:6
par(mar=c(5,6,4,2)+.1)
plot(infochange.means, x, main="Information Recall by Treatment Group\nThose Who Got 0/6-5/6 Pre Treatment",
     xlim=c(-1,2), xlab="Change in Recall (Post - Pre)", ylab="",
     axes=FALSE,
     col=c("black", "black", "seagreen", "black", "black", "black"),
     pch=19)
axis(1)
axis(2, at=1:length(labs), las=1, labels=labs)
arrows(infochange.means-1.96*infochange.se, x, infochange.means+1.96*infochange.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
#text(.55,6.10,"More Correct Recall Post-Treatment", cex=.75)
#text(-.55,6.10,"Less Correct Recall Post-Treatment", cex=.75)
box()


```

\FloatBarrier


### Economic Confidence 

The figure below shows the distribution of beliefs about the economy pre-treatment and post-treatment. Overall we see very little change, with a t test comparing the mean economic confidence index pre-treatment to post-treatment indicating no statistically significant difference. 

```{r echo=FALSE, eval=TRUE, fig.cap="Distribution of economic confidence index pre-treatment (solid) and post-treatment (dashed)."}

s2$econindex.pre <- NA
s2$econindex.pre <- (s2$econbetter.pre2 + s2$econconditions.pre2)/2

s2$econindex.post <- NA
s2$econindex.post <- (s2$econbetter.post2 + s2$econconditions.post2)/2

plot(density(s2$econindex.pre, na.rm=T), main="Distribution of Economic Confidence\nPre- and Post-Treatment", lwd=3, xlab="Economic Confidence Index", ylim=c(0,.8))
lines(density(s2$econindex.post, na.rm=T), lwd=3, lty=2)


```

\FloatBarrier

The following figures show the change in economic confidence among those who thought the economy was getting better pre-treatment and among those who thought it was getting worse pre-treatment. Among those who thought the economy was getting worse, the media did not lead them to update their beliefs at all. Instead, those who received information from Republican and Independent informants (as well as both ideal and non-ideal) updated their beliefs to thinking that the economy was getting better. This is consistent with the notion that information from Republicans is more likely to paint the economy in a more positive light than information from Democrats. Among those who thought the economy was getting better pre-treatment, we see that getting information from the media led them to update their beliefs to thinking it was getting worse, as did information from Democrats and Republicans (as well as ideal and non-ideal informants). That information from Republicans led to an update in this direction is surprising. 

```{r echo=FALSE, eval=TRUE}

s2sub <- s2[s2$econindex.pre<2.5,]

econindex.means <- c(mean(s2sub$econindex.change[s2sub$ideal.informant==0], na.rm=T),
                     mean(s2sub$econindex.change[s2sub$ideal.informant==1], na.rm=T),
                     mean(s2sub$econindex.change[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$econindex.change[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$econindex.change[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$econindex.change[s2sub$treatment=="socialrep"], na.rm=T))

econindex.se <- c(sd(s2sub$econindex.change[s2sub$ideal.informant==0], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$econindex.change)[1,])),
                  sd(s2sub$econindex.change[s2sub$ideal.informant==1], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$econindex.change)[2,])),
                  sd(s2sub$econindex.change[s2sub$treatment=="media"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["media",])),
                   sd(s2sub$econindex.change[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["socialdem",])),
                   sd(s2sub$econindex.change[s2sub$treatment=="socialind"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["socialind",])),
                   sd(s2sub$econindex.change[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["socialrep",])))
```


```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=6, fig.cap="Change in economic confidence index by treatment group. Horizontal lines represent 95 percent confidence intervals. The vertical line about zero represents no change in economic confidence."}

x <- 1:6
par(mfrow=c(1,1))
par(mar=c(5,6,4,2)+.1)
plot(econindex.means, x, main="Change in Economic Confidence Index\nAmong Those Who Thought the Economy was Getting Worse Pre-Treatment",
     xlim=c(-.25,.25), xlab="Change in Economic Confidence Index (Post - Pre)", ylab="",
     pch=19, col=c("black", "black", "seagreen", "black", "black", "black"),
     axes=FALSE, cex.main=.75)
axis(1)
axis(2, at=c(1,2,3,4,5,6), las=1, labels=c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant"))
arrows(econindex.means-1.96*econindex.se, x, econindex.means+1.96*econindex.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
text(.15,6.10,"Economy Better Post-Treatment", cex=.75)
text(-.15,6.10,"Economy Worse Post-Treatment", cex=.75)
box()

```

\FloatBarrier

```{r echo=FALSE, eval=TRUE}

s2sub <- s2[s2$econindex.pre>=2.5,]

econindex.means <- c(mean(s2sub$econindex.change[s2sub$ideal.informant==0], na.rm=T),
                     mean(s2sub$econindex.change[s2sub$ideal.informant==1], na.rm=T),
                     mean(s2sub$econindex.change[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$econindex.change[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$econindex.change[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$econindex.change[s2sub$treatment=="socialrep"], na.rm=T))

econindex.se <- c(sd(s2sub$econindex.change[s2sub$ideal.informant==0], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$econindex.change)[1,])),
                  sd(s2sub$econindex.change[s2sub$ideal.informant==1], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$econindex.change)[2,])),
                  sd(s2sub$econindex.change[s2sub$treatment=="media"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["media",])),
                   sd(s2sub$econindex.change[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["socialdem",])),
                   sd(s2sub$econindex.change[s2sub$treatment=="socialind"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["socialind",])),
                   sd(s2sub$econindex.change[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$econindex.change)["socialrep",])))
```


```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=6, fig.cap="Change in economic confidence index by treatment group. Horizontal lines represent 95 percent confidence intervals. The vertical line about zero represents no change in economic confidence."}

x <- 1:6
par(mfrow=c(1,1))
par(mar=c(5,6,4,2)+.1)
plot(econindex.means, x, main="Change in Economic Confidence Index\nAmong Those Who Thought the Economy was Getting Better Pre-Treatment",
     xlim=c(-.25,.25), xlab="Change in Economic Confidence Index (Post - Pre)", ylab="",
     pch=19, col=c("black", "black", "seagreen", "black", "black", "black"),
     axes=FALSE, cex.main=.75)
axis(1)
axis(2, at=c(1,2,3,4,5,6), las=1, labels=c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant"))
arrows(econindex.means-1.96*econindex.se, x, econindex.means+1.96*econindex.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
text(.15,6.10,"Economy Better Post-Treatment", cex=.75)
text(-.15,6.10,"Economy Worse Post-Treatment", cex=.75)
box()

```

\FloatBarrier

### Presidential Approval (overall)

The following figure shows the distribution of overall presidential approval before (solid) and after (dashed) treatment. Overall, respondents had low approval of President Trump and there was very little change after exposure to information. For example, 42.4 percent of respondents strongly disapproved pre-treatment, and 41.7 percent strongly disapproved post-treatment. 15.4 percent of respondents strongly approved pre-treatment and 16.2 percent strongly approved post-treatment. These minor changes reflect about ten respondents moving on the disapprove extreme and about five respondents moving on the approve extreme. There were, however, six respondents who moved from disapproving in some way (disapprove or strongly disapprove) to approving in some way (approve or strongly approve), but there was not movement in the opposite direction. 


```{r echo=FALSE, eval=TRUE, fig.cap="Distribution of presidential approval pre- (solid) and post-treatment (dashed)."}

plot(density(s2$trumpapprove.pre2, na.rm=T), main="Distribution of Presidential Approval\nPre- and Post-Treatment", lwd=3, xlab="Presidential Approval")
lines(density(s2$trumpapprove.post2, na.rm=T), lwd=3, lty=2)
```

\FloatBarrier

The following figures show the treatment effects for change in presidential approval among those who did not approve of Trump pre-treatment and those who did approve pre-treatment. Among those who disapproved pre-treatment, receiving information from a Republican led to an increase in presidential approval, as well as receiving information from a non-ideal informant. The non-ideal informant result is likely driven by Democrats who received information from a Republican. There is also suggestive evidence that receiving information from an Independent and from the media could lead to increased presidential approval, but receiving information from a Democrat lead to no updating at all. Among those who approved pre-treatment, no informational message led to a change in approval. 

```{r echo=FALSE, eval=TRUE}

s2sub <- s2[s2$trumpapprove.pre2<3,]

trumpapprove.change2.means <- c(mean(s2sub$trumpapprove.change2[s2sub$ideal.informant==0], na.rm=T),
                     mean(s2sub$trumpapprove.change2[s2sub$ideal.informant==1], na.rm=T),
                     mean(s2sub$trumpapprove.change2[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$trumpapprove.change2[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$trumpapprove.change2[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$trumpapprove.change2[s2sub$treatment=="socialrep"], na.rm=T))

trumpapprove.change2.se <- c(sd(s2sub$trumpapprove.change2[s2sub$ideal.informant==0], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpapprove.change2)[1,])),
                  sd(s2sub$trumpapprove.change2[s2sub$ideal.informant==1], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpapprove.change2)[2,])),
                  sd(s2sub$trumpapprove.change2[s2sub$treatment=="media"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["media",])),
                   sd(s2sub$trumpapprove.change2[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["socialdem",])),
                   sd(s2sub$trumpapprove.change2[s2sub$treatment=="socialind"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["socialind",])),
                   sd(s2sub$trumpapprove.change2[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["socialrep",])))


```


```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=6, fig.cap="Change in Trump approval by treatment group among those who did not approve of Trump pre-treatment. Horizontal lines represent 95 percent confidence intervals. The vertical line about zero represents no change."}

x <- 1:6
par(mfrow=c(1,1))
par(mar=c(5,6,4,2)+.1)
plot(trumpapprove.change2.means, x, main="Change in Trump Approval\nAmong Those Who Disapproved Pre-Treatment",
     xlim=c(-.25,.25), xlab="Change in Trump Approval (Post - Pre)", ylab="",
     pch=19, col=c("black", "black", "seagreen", "black", "black", "black"),
     axes=FALSE, cex.main=.75)
axis(1)
axis(2, at=c(1,2,3,4,5,6), las=1, labels=c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant"))
arrows(trumpapprove.change2.means-1.96*trumpapprove.change2.se, x, trumpapprove.change2.means+1.96*trumpapprove.change2.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
text(.15,6.10,"Greater Approval Post-Treatment", cex=.75)
text(-.15,6.10,"Less Approval Post-Treatment", cex=.75)
box()

```

\FloatBarrier

```{r echo=FALSE, eval=TRUE}

s2sub <- s2[s2$trumpapprove.pre2>2,]


trumpapprove.change2.means <- c(mean(s2sub$trumpapprove.change2[s2sub$ideal.informant==0], na.rm=T),
                     mean(s2sub$trumpapprove.change2[s2sub$ideal.informant==1], na.rm=T),
                     mean(s2sub$trumpapprove.change2[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$trumpapprove.change2[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$trumpapprove.change2[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$trumpapprove.change2[s2sub$treatment=="socialrep"], na.rm=T))

trumpapprove.change2.se <- c(sd(s2sub$trumpapprove.change2[s2sub$ideal.informant==0], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpapprove.change2)[1,])),
                  sd(s2sub$trumpapprove.change2[s2sub$ideal.informant==1], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpapprove.change2)[2,])),
                  sd(s2sub$trumpapprove.change2[s2sub$treatment=="media"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["media",])),
                   sd(s2sub$trumpapprove.change2[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["socialdem",])),
                   sd(s2sub$trumpapprove.change2[s2sub$treatment=="socialind"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["socialind",])),
                   sd(s2sub$trumpapprove.change2[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpapprove.change2)["socialrep",])))
```


```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=6, fig.cap="Change in Trump approval by treatment group among those who approved of Trump pre-treatment. Horizontal lines represent 95 percent confidence intervals. The vertical line about zero represents no change."}

x <- 1:6
par(mfrow=c(1,1))
par(mar=c(5,6,4,2)+.1)
plot(trumpapprove.change2.means, x, main="Change in Trump Approval\nAmong Those Who Approved Pre-Treatment",
     xlim=c(-.25,.25), xlab="Change in Trump Approval (Post - Pre)", ylab="",
     pch=19, col=c("black", "black", "seagreen", "black", "black", "black"),
     axes=FALSE, cex.main=.75)
axis(1)
axis(2, at=c(1,2,3,4,5,6), las=1, labels=c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant"))
arrows(trumpapprove.change2.means-1.96*trumpapprove.change2.se, x, trumpapprove.change2.means+1.96*trumpapprove.change2.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
text(.15,6.10,"Greater Approval Post-Treatment", cex=.75)
text(-.15,6.10,"Less Approval Post-Treatment", cex=.75)
box()

```

\FloatBarrier

### Approval of How Trump is Handling the Economy

The figure below shows the distribution of approval of how Trump is handling the economy pre-treatment and post-treatment. The results show virtually no change after exposure to information. 

```{r echo=FALSE, eval=TRUE, fig.cap="Distribution of approval of how Trump is handling the economy pre- (solid) and post- (dashed) treatment."}

plot(density(s2$trumpeconapprove.pre2, na.rm=T), main="Distribution of Approval of Trump Handling the Economy\nPre- and Post-Treatment", xlab="Approval of Trump Handling Economy", lwd=3, ylim=c(0,.6))
lines(density(s2$trumpeconapprove.post2, na.rm=T), lwd=3, lty=2)

```

\FloatBarrier

The following figures show the treatment effects for change in approval of how Trump is handling the economy among those who approved pre-treatment and disapproved pre-treatment. Among those who disapproved pre-treatment, no informational treatments led to changes in approval. However, among those who approved pre-treatment (n=414), receiving information from the media or an Independent led to a decrease in approval. 

```{r echo=FALSE, eval=TRUE}

s2sub <- s2[s2$trumpeconapprove.pre2<3,]

trumpeconapprove.change2.means <- c(mean(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==0], na.rm=T),
                     mean(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==1], na.rm=T),
                     mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialrep"], na.rm=T))

trumpeconapprove.change2.se <- c(sd(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==0], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpeconapprove.change2)[1,])),
                  sd(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==1], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpeconapprove.change2)[2,])),
                  sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="media"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["media",])),
                   sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["socialdem",])),
                   sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialind"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["socialind",])),
                   sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["socialrep",])))
```


```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=6, fig.cap="Change in approval of Trump's handling the economy by treatment group among those who did not approve pre-treatment. Horizontal lines represent 95 percent confidence intervals. The vertical line about zero represents no change."}

x <- 1:6
par(mfrow=c(1,1))
par(mar=c(5,6,4,2)+.1)
plot(trumpeconapprove.change2.means, x, main="Change in Trump Approval (Handling Economy)\nAmong Those Who Disapproved Pre-Treatment",
     xlim=c(-.25,.25), xlab="Change in Trump Approval (Post - Pre)", ylab="",
     pch=19, col=c("black", "black", "seagreen", "black", "black", "black"),
     axes=FALSE, cex.main=.75)
axis(1)
axis(2, at=c(1,2,3,4,5,6), las=1, labels=c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant"))
arrows(trumpeconapprove.change2.means-1.96*trumpeconapprove.change2.se, x, trumpeconapprove.change2.means+1.96*trumpeconapprove.change2.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
text(.15,6.10,"Greater Approval Post-Treatment", cex=.75)
text(-.15,6.10,"Less Approval Post-Treatment", cex=.75)
box()

```

\FloatBarrier

```{r echo=FALSE, eval=TRUE}

s2sub <- s2[s2$trumpeconapprove.pre2>2,]

trumpeconapprove.change2.means <- c(mean(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==0], na.rm=T),
                     mean(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==1], na.rm=T),
                     mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="media"], na.rm=T),
                mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialdem"], na.rm=T),
                mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialind"], na.rm=T),
                mean(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialrep"], na.rm=T))

trumpeconapprove.change2.se <- c(sd(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==0], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpeconapprove.change2)[1,])),
                  sd(s2sub$trumpeconapprove.change2[s2sub$ideal.informant==1], na.rm=T)/sqrt(sum(table(s2sub$ideal.informant, s2sub$trumpeconapprove.change2)[2,])),
                  sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="media"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["media",])),
                   sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialdem"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["socialdem",])),
                   sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialind"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["socialind",])),
                   sd(s2sub$trumpeconapprove.change2[s2sub$treatment=="socialrep"], na.rm=T)/sqrt(sum(table(s2sub$treatment, s2sub$trumpeconapprove.change2)["socialrep",])))
```


```{r echo=FALSE, eval=TRUE, fig.height=5, fig.width=6, fig.cap="Change in approval of Trump's handling the economy by treatment group among those who approved pre-treatment. Horizontal lines represent 95 percent confidence intervals. The vertical line about zero represents no change."}

x <- 1:6
par(mfrow=c(1,1))
par(mar=c(5,6,4,2)+.1)
plot(trumpeconapprove.change2.means, x, main="Change in Trump Approval (Handling Economy)\nAmong Those Who Approved Pre-Treatment",
     xlim=c(-.25,.25), xlab="Change in Trump Approval (Post - Pre)", ylab="",
     pch=19, col=c("black", "black", "seagreen", "black", "black", "black"),
     axes=FALSE, cex.main=.75)
axis(1)
axis(2, at=c(1,2,3,4,5,6), las=1, labels=c("Non-Ideal\nInformant", "Ideal\nInformant", "Media", "Democrat\nInformant", "Independent\nInformant", "Republican\nInformant"))
arrows(trumpeconapprove.change2.means-1.96*trumpeconapprove.change2.se, x, trumpeconapprove.change2.means+1.96*trumpeconapprove.change2.se, x, angle=90, code=.5, col=c("black", "black", "seagreen", "black", "black", "black"))
abline(v=0)
text(.15,6.10,"Greater Approval Post-Treatment", cex=.75)
text(-.15,6.10,"Less Approval Post-Treatment", cex=.75)
box()

```

\FloatBarrier


## Survey Instrument

### Consent 

The general nature of this study entitled "People and Attitudes" conducted by [NAME REDACTED] has been explained to me. I understand that I will be asked to answer a series of questions about myself and my behavior. My participation in this study should take a total of about 20 minutes. Participation in this study may involve some risks of boredom, frustratino, anxiety, and anger. To minimize these risks you may choose to not answer any of the questions or to stop the survey at any time. I understand anonymity will be preserved and that my name will not be associated with any results of this study.I know that I may refuse to answer any question asked and that I may discontinue participation at any time.I also understand that any payment for participation will not be affected by my responses or by my exercising any of my rights. Potential risks resulting from my participation in this project have been described to me. I am aware that I may report dissatisfactions with any aspect of this study to the Chair of the Protection of Human Subjects Committee, [NAME REDACTED, CONTACT INFORMATION REDACTED]. I am aware that I must be at leaset 18 years of age to participate. 
\begin{itemize}
  \item I consent to participate in this study 
  \item I do not consent to participate in this study 
\end{itemize}

### Pre-treatment Questions

\begin{itemize}
  \item We'd first like to ask you a few questions about the U.S. economy. 
  \item PAGE BREAK
  \item We'd like to ask you a few questions about the current state of the US economy. Please indicate whether you think each of the following statements are TRUE or FALSE. 
    \begin{itemize}
    \item The US economy grew in the first quarter of 2017
    \item GDP grew at the fastest rate since the second quarter of 2016 in the US
    \item GDP in January-March tends to over-perform relative to the rest of the year
    \item In the first quarte rof 2017, GDP grew at a slower rate than the Trump administration's target
    \item Since 2000, the US economy has grown at an average rate of 0.5\%
    \item Consumer spending accounts for less than 1/4 of US economic activity
    \end{itemize}
  \item PAGE BREAK
  \item Below are your responses to the previous questions. How confident are you that you answered each question correctly? \textit{Statements from the previous question were repeated, along with the True or False answer selected by the participant. The response options included: not very confident, somewhat confident, confident, very confident}
  \item PAGE BREAK
  \item Right now do you think that economic conditions in the country as a whole are getting better or getting worse? \textit{Getting better, getting worse, staying the same, don't know}
  \item How would you rate economic conditions in this country today? \textit{Excellent, good, only fair, poor, don't know}
  \item PAGE BREAK
  \item Do you approve or disapprove of the way Donald Trump is handling his job as president? \textit{Strongly approve, approve, disapprove, strongly disapprove, no opinion}
  \item Do you approve or disapprove of the way Donald Trump is handling the economy? \textit{Strongly approve, approve, disapprove, strongly disapprove, no opinion}
  \item PAGE BREAK
  \item When thinking politically, what do you usually consider yourself--a Republican, a Democrat, an Independent, or what? \textit{Democrat, Republican, Independent}
  \item PAGE BREAK
  \item IF INDEPENDENT: Do you lean more towards the Republican Party or the Democratic Party? \textit{Democratic Party, Republican Party, Neither}
  \item IF NOT INDEPENDENT: Do you consider yourself a strong [insert selected partisanship -- Republican or Democrat] or not a very strong [insert selected partisanship -- Republican or Democrat]? \textit{Strong [insert selected partisanship -- Republican or Democrat], Not very strong [insert selected partisanship -- Republican or Democrat]}
\end{itemize}

### Distractor Questions 

\begin{itemize}
  \item PAGE BREAK
  \item People get information about politics, candidates, and elections from many different sources. How about you? About what percent of the information you have about politics comes from each source? \textit{Television, Radio, Newspaper articles (online or in print) that I seek out on my own, Newspaper articles (online or in print) that are referred to me by a friend, Conversations with other people (online or face to face)}
  \item Other questions about initiating conversations about politics with others as part of other studies. 
  \item Other questions about social polarization and social distancing as part of other studies.
  \item Other questions about social media use as part of other studies.
\end{itemize}

### Informational Treatments 

\begin{itemize}
  \item Now we are going to present you with some information about economic performance in the US in the first quarter of 2017. 
  \item PAGE BREAK 
  \item \textbf{MEDIA TREATMENT}
    \begin{itemize}
    \item Please read the following information about the economic performance in the US in the first quarter of 2017. This information comes from a news article published by Reuters. 
    \item SHOWN NEWS ARTICLE
    \end{itemize}
  \item \textbf{DEMOCRAT INFORMANT TREATMENT}
    \begin{itemize}
    \item Please read the following information that comes from a Democrat who read an article about the US economic performance in the first quarter of 2017 published by Reuters. 
    \item SHOWN MESSAGE WRITTEN BY A DEMOCRAT, WRITTEN TO SOMEONE OF THE RESPONDENT'S PARTISANSHIP
    \end{itemize}
  \item \textbf{REPUBLICAN INFORMANT TREATMENT}
    \begin{itemize}
    \item Please read the following information that comes from a Republican who read an article about the US economic performance in the first quarter of 2017 published by Reuters. 
    \item SHOWN MESSAGE WRITTEN BY A Republican, WRITTEN TO SOMEONE OF THE RESPONDENT'S PARTISANSHIP
    \end{itemize}
  \item \textbf{INDEPENDENT INFORMANT TREATMENT}
    \begin{itemize}
    \item Please read the following information that comes from a Independent who read an article about the US economic performance in the first quarter of 2017 published by Reuters. 
    \item SHOWN MESSAGE WRITTEN BY A INDEPENDENT, WRITTEN TO SOMEONE OF THE RESPONDENT'S PARTISANSHIP
    \end{itemize}
\end{itemize}

### Post-treatment Questions 

\begin{itemize}
  \item PAGE BREAK
  \item We'd like to ask you a few questions about the current state of the US economy. Please indicate whether you think each of the following statements are TRUE or FALSE. 
    \begin{itemize}
    \item The US economy grew in the first quarter of 2017
    \item GDP grew at the fastest rate since the second quarter of 2016 in the US
    \item GDP in January-March tends to over-perform relative to the rest of the year
    \item In the first quarte rof 2017, GDP grew at a slower rate than the Trump administration's target
    \item Since 2000, the US economy has grown at an average rate of 0.5\%
    \item Consumer spending accounts for less than 1/4 of US economic activity
    \end{itemize}
  \item PAGE BREAK
  \item Below are your responses to the previous questions. How confident are you that you answered each question correctly? \textit{Statements from the previous question were repeated, along with the True or False answer selected by the participant. The response options included: not very confident, somewhat confident, confident, very confident}
  \item PAGE BREAK
  \item Right now do you think that economic conditions in the country as a whole are getting better or getting worse? \textit{Getting better, getting worse, staying the same, don't know}
  \item How would you rate economic conditions in this country today? \textit{Excellent, good, only fair, poor, don't know}
  \item PAGE BREAK
  \item Do you approve or disapprove of the way Donald Trump is handling his job as president? \textit{Strongly approve, approve, disapprove, strongly disapprove, no opinion}
  \item Do you approve or disapprove of the way Donald Trump is handling the economy? \textit{Strongly approve, approve, disapprove, strongly disapprove, no opinion}
  \item PAGE BREAK
  \item When thinking politically, what do you usually consider yourself--a Republican, a Democrat, an Independent, or what? \textit{Democrat, Republican, Independent}
  \item PAGE BREAK
  \item IF INDEPENDENT: Do you lean more towards the Republican Party or the Democratic Party? \textit{Democratic Party, Republican Party, Neither}
  \item IF NOT INDEPENDENT: Do you consider yourself a strong [insert selected partisanship -- Republican or Democrat] or not a very strong [insert selected partisanship -- Republican or Democrat]? \textit{Strong [insert selected partisanship -- Republican or Democrat], Not very strong [insert selected partisanship -- Republican or Democrat]}
\end{itemize}

# Message Treatments

Participants were randomly assigned to read a socially transcribed message or a news article. The messages were stored in individual questions within treatment blocks within qualtrics. So, each message has a unique question that will take the value of 1 if the participant was exposed to that question / message and NA otherwise. In this section of the codebook, I include the full text of the messages with the corresponding question number. 

## Social Independent Treatment --- Independent Participants

Participants who identify as Independents who were randomly assigned to read a message written by an Independent. Thus, these messages were written by Independents to Independents. All messages started with "Please read the following information that comes from an Independent who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q78 --- The economy is having a lot of issues right now.

* Q79 --- The economy is reported as growing and doing better during the first quarter of 2017

* Q80 --- National gross domestic product has grown by only 1.1-1.4 percent in the first quarter which is pretty low even though trump and his administration were hoping to get it to grow by 3\% compared to last year.

* Q81 --- Most economic aspects of the country are improving, at least slightly. The targeted growth rate of 3\% yearly is challenging, but there's some growth.

* Q82 --- Growth is slow, but there's growth. I don't really care about the economy, at least, not in statistics and numbers.

* Q83 --- The eocnomy is still going despite the government not actually doing anything. There has been little action in the Trump administration. 

* Q84 --- Even though Trump and his administration along with the Republican majority congress don't know their left hand from their right hand, business seems to be on the upswing. For some reason the economy likes Republicans. Probably because they give enormous tax breaks to the rich to make them richer while taking away from every one else to pay for it.

* Q85 --- The economy is doing marginally better than initially thought. But we are still way to dependent on consuming.

* Q86 --- The president thinks he can actually make a change for this company, but I think he's just doing more harm than good.

* Q87 --- The article talkeda bout economic growth and since Trump took office things are not as rosy as predicted. The GDP increased a bit more than expected due to consumer spending. Trumps promises of cuts and economic promises have yet to get off the ground, clearly written by a liberal. Also business spending was up.

* Q88 --- I read that in the macro sense, economic growth in the country overall is still relatively slow, but not as bad as was forecast. 

* Q89 --- I feel like the economy is on an uptick. Although President Trump has not been able to meet his agenda as he wanted, the economy is doing better. 

* Q90 --- While I - - - - - - - hate Donald Trump it seems that he is actually doing some good. He is still a moron but he appears to be getting numbers up so thats good.

* Q91 --- It takes time to see lasting effects of presidential changes to economic policy.

* Q92 --- The article talked about the growth of the economy. The economy is growing. It is not at target set by the government. Policies are slowly taking in effect.

* Q93 --- Im surprised about how tremendously well the economy is starting to improve in the U.S. Even if it is getting better slowly is still a positive aspect. Hopefully it will continue and open up more opportunities and resources required for people to live more successful lives. 

* Q94 --- The US's GDP still looks OK this year.

* Q95 --- The economy is growing slowly. Trump has not completed many of his tasks in the first 5 months.

* Q96 --- The economy, as measured by the GDP is still more or less not really growing. It is not far off from the same middle of 1-2 percent of 2016. Trump is looking for 3 percent like the 1990s. Other business metrics are mixed and negative. 

* Q97 --- The report I read said that although the actual GDP was up sligthly form the estimate and forecasts, it still was low. Trump wants to see a 3\% growth rate but currently it's only 1.4\%. Overall there was a mix of good and bad elements, like exports were up, but so were inventories. So 3\% isn't likely to happen anytime soon.

* Q98 --- There was higher consumer spending and a bigger jump in exports during the first quarter in the U.S. It was up 1.4 percent.

* Q99 --- The GDP has not done as well as the current president promised. His economic program of tax cuts, regulatory rollbacks and infrastructure spending has yet to get off the ground since the 5 months of his presidency. Our economy has not increased so far this year.

* Q100 --- Our country is in danger, nothing is being done. The numbers aren't even accurate!

* Q101 --- Things seemed to be looking up with the stock market. There has been in increase and thats usually a good thing. 

* Q102 --- The US economy was not as bad as estimated. Consumers spent more than initially thought, making the depression less than expected.

* Q103 --- GDP rate increased at 1.4 last quarter, and probably would remain constant at 1.2 next.

* Q104 --- I just read an article on the revised numbers of GDP for the last quarter. All Icould get out of it was that it wasn;t as bad as they initially thoought. Beyond that, I could notget anything out the articale. It felt like I was reading mumbo jumbo. 

* Q105 --- The economy is doing better than those who think of these things, thought it would. I surprise, especially exports and such being up. Lots of trade with other countries, even during the current administration.

* Q106 --- The article said our economic situation is good just not as good as we were promised.

* Q107 --- Economic growth in the us continues to underperform compared to historical standards.

* Q108 --- It appears, based on the article, that our country's economic growth has somewhat improved rather than the forecasts have predicted would occur.

## Social Independent Treatment --- Republican Participants

Participants who identify as Republicans who were randomly assigned to read a message written by an Independent. Thus, these messages were written by Independents to Republicans. All messages started with "Please read the following information that comes from an Independent who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q73 --- Despite many criticisms the economy si growing at a faster pace than expected.

* Q74 --- I think if you want spending to increase, there would have to be more help for the middle class. We can't spend more if we're barely scraping by to pay our bills. 

* Q75 --- The economy grew 1.6\% last quarter.

* Q76 --- The economy is boosting in the first quarter under president Trump, but not as much as expected. The economy slowed more than expected due to higher consumer spending and more exports. 

* Q77 --- Although I don't agree with much of what trump says, he's not doing that bad of a job.

* Q109 --- The growth is not going very well and it doesn't look like the President is doing a very good job of holding up his end.

* Q110 --- I've read a really interesting article the other day. Did you know that gross domestic product increased at a 1.4 annual rate instead of 1.2 last month? It is still the slowest growth rate but during January - march it tends to under perform. Also, president donald trump's economic program of tax cuts still hasn't been done, despite him being in office for five months. What's the problem?

* Q111 --- Well, I have very little idea what I just read. It seems that the GDP grew at a slightly faster rate than expected but other economic signs are down. Growth is nowhere near what it was in the 1990s, and Trumps plan to raise (something) to 3 percent has not panned out.

* Q112 --- Republicans cannot not even be successful with their new president with getting a clear budget cut done.

* Q113 --- The new administration said that they are targeting 3\% yearly growth to get the economy going.

* Q114 --- The target goal of growth is a bit steep but hopefully everything works out for the best.

* Q115 --- While Donald Trump is saying he is the most successful president since Abraham Lincoln in truth his economic plan hasn't even really began nevertheless have any ability to have any outcome yet. There is no healthplan. Nothing has occurred. 

* Q116 --- Things are being rewritten to show increases.

* Q117 --- It appears the Trump administration is having a hard time implementing all the policies and things he said he would. It must be very disappointing to you as Republican especially since it is Republican controlled congress as well! Republicans have the presidency and control both houses of congress and they still can't even come up with something to repeal and replace Obama Care Affordable Care Act let alone fix the economy! Must be very discouraging indeed.

* Q118 --- While I'm willing to give Trump a chance, I would be monitoring the economy's progress carefully, especially since it seems to affect most, if not all of us, in some way, shape or form.

* Q119 --- The economy is currently at it's slowest for quite some time.

* Q120 --- The data is conflicting because there are both tentatively positive results withs omewhat negative results, so it's a mixed bag. There are things the current administration have set as goals which not only are not realized but may in fact be unrealistic: the drive for 3\% is ridiculous in the light that such a thing hasn't been seen since the 90's and even then it was lower. That was 30 years ago. 

* Q121 --- It is difficult to make an accurate assessment of the Trump administration economic policies 5 months into his administration. There are a great deal of numbers that give conflicting accounts of the current economy but the bottom line is the economy is growing, exports are increasing, inventories are down and the unemployment rate is staying consistent. IT will be interesting if "America First" works when looked at over a reasonable amount of time.

* Q122 --- The economy has picked up slightly but not at the rate that the President was hoping for. The overall spending in the economy by the consumer is up and that is one of the major reasons why there was a slight uptick in the economy this last quarter. We still had a lot of work ahead of ourselves before the president could reach his goal and that this was only going to happen if we agreed to work together. 

* Q123 --- Its been 5 months since the new President's presidency and his economic program of tax cuts, regulatory rollbacks, and infrastructure spending has not made an impact. 

* Q124 --- Focus on actions and make laws that would actually benefit both the businesses and consumers. The consumer holds the key to economic growth, and suppressing them will inflect upon the economic growth of the country. Stop buying into Trump's antics about how the economy is growing and foster real change in the economy. 

* Q125 --- The U.S. economy is growth because of consumer spending. Therefore, policies that encourage business development, reduce taxes, reduce regulations, and reduce dependence on foreign imports will stimulat ethe overall economy. Although generally the economy is improving, it will be difficult to meet the goals for GDP growth that the president is aiming to achieve. Nonetheless, the economy is increasing and therefore the current economic outlook is stable and improving compared less than a decade ago during the 2008 mortgage and stock market crisis.

* Q126 --- GDP growth was slightly higher than expected last quarter. Consumer confidence is up. This is continuing the same trend at the tail end of the Obama administration. None of president's Trump's legislative policies have yet to get off the ground. 

* Q127 --- You know, the most recent reporting on the U.S. economy showed a bit of an improvement owing to some unexpectedly moderate rise in consumer psending \& a bigger jump in exports. This also included a nice showing on Gross domestic product according to the Commerce Department final assessment on prior first quarter estimates, as Thursday's reporting says. A sustained average of 3 percent growth has not been seen since the 1990s. Since 2000, the U.S. economy has grown at an average 2 percent rate. The Trump administration's stated that it is still expecting a target of swiftly boosting U.S. growth to 3 percent. We shall see how things develop going foreword, given President Donald Trump's economic program of tax cuts, regulatory rollbacks, and infrastructure spending. 

## Social Independent Treatment --- Democrat Participants

Participants who identify as Democrats who were randomly assigned to read a message written by an Independent. Thus, these messages were written by Independents to Democrats. All messages started with "Please read the following information that comes from an Independent who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q68 --- The economy is improving as shown in the first quarte r

* Q69 --- Despite everything that's happening, the economy is doing well and will continue to do so. 

* Q70 --- I just read an article that I barely understand. It said something about the US economy has slowed a lot. It also said that the Trump administration's plan of boosting the US economy to 3 percent was something that would be tough to do and that it has not been 3 percent since the 90s. Most of it was technical mumbo-jumbo but it said the US economy was bad. 

* Q71 --- trump sucks and the economy is growing too quickly.

* Q72 --- It seems that earnings in spending and similar transactions are better than slightly predicted and trump has nothing to do with it. However values are still some of the lowest we have seen in a while.

* Q128 --- Some reported data about the economy seems to indicate that things are improving. 

* Q129 --- I find it troubling that we live in these time's. I think some thing should change immediately. 

* Q130 --- I just read an article that seemed to indicate the economy is doing a little better than expected and it's giving me a little hope about the future.

* Q131 --- US first quarter growth is slow. President Trump's policies have yet to take affect on the macro economy. 

* Q132 --- Since the year started, our nation has earned 1.4\% than our total worth so far. That is called GDP. We are making good pace and making lots of money.

* Q133 --- The presidents plan to improve the economy is taking longer than expected.

* Q134 --- Economic indicators are looking more positive since Trump was elected, despite the fact his economic program of tax cuts, regulatory rollbacks and infrastructure spending has yet to get off the ground.

* Q135 --- President Trump had made plans of the cutting of spending in several different areas to increase GDP growth overall. The growth had been successful at 1.4\% but is noted that it is one of the slowest growths to occur. 

* Q136 --- The economy is improving at a faster rate, but it's still almost the lowest percentage since the second quarter of last year.

* Q137 --- It would be nice if we could get tax reform done this year, it's nice to see somebody making some kind of effort to rollback the encroachment of the adnimistrative state and reign in excessive regulation by un-elected bureaucrats, and I'm hopeful the two major parties can find a way to work together on much-needed and over-delayed nationwide infrastructure investments. 

* Q138 --- The decrease in the economy. The average of 3\% growth hasn't been seen since the 1990's. I was upset to hear this and it needs to change and more opportunities need to become available for the American people and stop relying so much on outside sources.

* Q139 --- I feel that reducing costs can be a good thing. I feel this is leading down the right path. 

* Q140 --- The article is stating that there is growth, but it is small and there is a decline in the slowing down of the GDP. Even though the President is having trouble getting some programs off the ground There has been some growth. There has been an increase in consumer spending and exporting of goods.

* Q141 --- Trump has promised an economic increase of 3\% but has yet to deliver on that.

* Q142 --- Recent GDP growth has exceeded projections but still remains very low at an annual rate of 1.4\%. This is still pretty far off of the target of 3\% that the current administration is targeting. 

* Q143 --- The GDP isn't as terrible as predicted, though it's still not good. People still aren't buying anything, which, you know, will happen when people don't have enough money to subsist on.

* Q144 --- The Trump administration plan for growing the economy at three percent each year is looking more improbable each quarter. Spending and growth have both shown an increase, but are still far from the averages shown during the last couple years. 

* Q145 --- The US growth rate has improved more than previously thought under President Trump.

## Social Republican Treatment --- Independent Participants

Participants who identify as Independents who were randomly assigned to read a message written by a Republican. Thus, these messages were written by Republicans to Independents. All messages started with "Please read the following information that comes from a Republican who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q63 --- The economic growth our wonderful president promised us and is fighting hard for isnt coming as fast as the rate is was supposed to have grown.

* Q64 --- Its a ever changing thing when it comes to the stock market.

* Q65 --- The economy slowed a bit in the first quarter but there were still signs that the economy is growing. This is without all of the legislation that is to come on taxes and spending on infrastructure.

* Q66 --- Although the economy is not improving at the rate that it has in the recent past, this past quarter's performance was better than projected.

* Q67 --- I don't think the economy is performing as well as our newly elected President would like us to believe. An article came out with numbers that have shown a consistent lack of performance in our economy. The data just isn't there to support an expansion of growth at the rate we are being told. For instance, if memory serves me correct business inventories are down to 2.6 billion instead of 4.3 billion that was reported the previous month. I'm just an average woman but the information and data collected is not supporting the big growth that we were told would happen. 

* Q146 --- I recently read that since the year 2000 our economy has grown at a 2 percent rate every year. Trump is hoping to raise that to 3 percent. The growth of our economy has slowed down. 

* Q147 --- The economy seems good statistically but I'm a little concerned about geo-politics getting in the way. We both agree that so called president Trump is a hot head and I think we'll likely see some of military action either in the middle east or east Asia as a distraction from his Russian collusion investigation. This will bring the markets down and productivity will be stifled so the current data is very fragile. 

* Q148 --- The rate has went up to 1.4 and seems to be doing better in growths aspects.

* Q149 --- Due to consumer spending, the US Economy had a boost in the first quarter. It was 1.4 instead of 1.2. Trump's economic program is not to thank because it hasn't gone into effect yet. Exports raised to 7.0.

* Q150 --- The rise in economic growth was higher than predicted but it still was the lowest it has been in the last 5 years.

* Q151 --- Economic growth in the first quarter was slightly higher than previously reported, but it's still slower than it has been in the past several years. The increase over the previous estimate was due to stronger consumer spending and exports.

* Q152 --- Even though economists predicted GDP to remain unchanged, it actually increased over the past few months. The Trump presidency isn't showing a significant growth rate in the economy yet but it looks to be pretty promising. 

* Q153 --- The Economy has been on a slow climb. President Trump had planned to boost by 3\% a year but is currently only around half of that right now. His plan has not gotten off the ground in the 5 months he's been in office and we haven't seen an increase of that size since the 1990's. 

* Q154 --- The eocnomy may not be growing as fast as we would like for it to but at least it is on the rise which is what we need.

* Q155 --- The US economy was better in the first quarter than expected due to higher consumer spending and in increase ine xports. Good but still the slowest quarter in a year. We have a long way to go to meet Trumps expectations of 3\% growth which hasn't been seen since the 1990's. 

* Q156 --- Within the first quarter the economic growth exceeded expectations. We still have not hit President Trump's target of a 3\% growth, however, a 1.6\% increase is making progress.

* Q157 --- GDP growth was revised to be a little better than initially thought for the first quarter. The economy still has a way to go to reach a GDP growth of 3\%. Trumps pro business ideas have yet to take effect. 

* Q158 --- The were so high spots in the final report from the Commerce Department, but not the three percent growth for which President Trump had hoped. There have been fits and starts in exports and economic growth and other area which are promised, but still not enough to meet the promised level. 

* Q159 --- Every tiem there is a new president the economy takes a hit. It takes time for change to happen. No matter how fast a candidate says it will happen it never does. Just because people dont like trump doesn't mean he wont deliver on his promises. 

* Q160 --- It looks as though the economy is doing fairly well. Consumer spending is up and this is driving the upswing in other things. Jobs are on the rise by a small rise. Businesses are seeing tax relief which allows them to hire. 

* Q161 --- The U.S economic performance in the first quarter of 2017 and that the GDP growth revised up to 1.4 percent.

* Q162 --- The economy, while stable, has slowed down more than expected. Given the political climate, this is a surprise as promsies were made. However, five months isn't really enough to see the result of any effective change. The numbers from the first quarter are admittedly lackluster, but I think the economy for the fiscal year shoul dbe evaluated as a whole. I'm for a wait-and-see approach, which I think the article was advocatign without saying as much. 

* Q163 --- Seems like our economy is growing slightly, but not as much as President Trump predicated, and overall the growth is a bit stilted and slow. 

* Q164 --- Its surprising that growth has been such a challenge. And profits fell after tax in the first quarter after rising the last 3 months. 

* Q165 --- The economy isn't doing as well as it use to and growth has slowed down significantly. This may be due to trump but things will probably get better in the future. 

* Q166 --- GDP rose 1.6 percent instead of 1.4 percent the government said. I suppose that is a bit better. 

* Q167 --- The GDP was revised to show a 1.4 increase, up from 1.2.

* Q168 --- US economy is doing slightly better, then expexted , in the first quarter of 2017. GPD was growing 1.4\% , while 1.2 \% was expected. The goal ,President Trump stated before (3\%) remains a challenge.

* Q169 --- The article I read stated that GDP was still fairly low for the first quarter of 2017, and has not gotten up to 3\% as offered by the Trump administration. However forecasters estimate the 2nd quarter GDP to be 2.9\%. Retail sales figures slumped in the first quarter of 2017 and there are cyclical factors in the governments' analysis that causes the first quarter GDP to always be a little deflated compared to other quarters of the year. 

* Q170 --- The GDP increased 1.4 percent annual rate when it was expected to go up 1.2 percent. That is still the lowest since last year. Trump administration's stated target of swiftly booting U.S. growth to 3 percent remains a challenge. President Donald Trump's economic program of tax cuts, regulatory rollbacks and infrastructure spending has yet to get off the ground five months into his presidency. 

* Q171 --- There have been decreases in fiscal years and that there are methods that are in play to figure out how to increase funds.

* Q172 --- The government growth forecast for this first part of the year seems to be flawed. The president wants a growth of over 3\%, but it stands at 1.4\% for the first quarter. Exports were at 7\%, but the promises of the campaign have yet to come to fruition. Infrastructure, tax reductions, and increased domestic output arenot showing up in this current quarter. Perhaps when the government gets its information straight, and the population starts to spend more we will see a better economy. 

* Q173 --- For the most part, it would seem that our economy is doing well, and growing!

* Q174 --- According to the article, we are seeing some growth in the GDP in the first quarter of this year. President Trump is giving tax cuts for major businesses, which have been exporting a lot more. Consumers have also affected the GDP by spending more than usual. Although their is some growth, it is much smaller than what the US has done in the best and these trade deals could prove to be valuable for the future of our economy. 

* Q175 --- The US economy did not grow as fast as predicted

* Q176 --- Though growth has happened since Trump became president, it's not sustainable. He can't do the impossible, and I don't think he would be able to do it. 

* Q177 --- The article states that the economy is not improving and has been stagnant over a large number of years.

* Q178 --- It seems that there has been a slight up-tick in economic numbers in the first quarter of 2017. It isn't as robust as was hoped, but it still is enough, to be optimistic about the future of the GDP. Analysts are saying that it might never be as robust as it was prior to 1990 when it was pretty steadily at or above 3\%, but it could get back into the 2\% range.

* Q179 --- For someone with a financial background, who enjoys reading the business section of the newspaper, I would recommend to an Independent not to read this article. The article presents figure after figure after figure. It makes your head spin. Estimates are purely that estimtes. Reported growth of the GNP or any other financial data can be too easily manipulated to report what one wants to prove. I would tell an Independent to look beyond this article and study the stock market, especially the S\&P 500. Consumer confidence is abounding despite the fact that the President's favorability rating is low. Go figure! 

* Q180 --- I'm not sure about the article. It seemed to emphasize negative aspects of an uptick in growth. I get that the growth numbers seem beind the goal of 3\%. It takes time to change, but wonder if 3\% might be too ambitious. still it's growing and I don't understand the "down" emphasis of the article. 

* Q181 --- The gdp rose slightly but is still below 2\% and getting higher is going to be a challenge.

* Q182 --- You have to look at the growth in a few positive ways. If corporations are doing better than they will hire more employees. They will also give better salaries to the employees that they already have. People are able to take care of themselves and their families. This is positive growth and in the few months that Trump has been in office he has done more than the previous President did. 

## Social Republican Treatment --- Republican Participants

Participants who identify as Republicans who were randomly assigned to read a message written by a Republican. Thus, these messages were written by Republicans to Republicans. All messages started with "Please read the following information that comes from a Republican who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q58 --- Business growth has not been as projected the past few month. Though the United States economic growth rate has stopped reaching the peak which it achieved in the 1990s. Most key indicator including housing, manufacturing, export..shows there is a slow but steady increase, but not at the level that has been as propserous as the past. 

* Q59 --- Economic growth is still slower than it used to be, but still seems to be trending upwards. It seems like consumers are spending more than anticipated, and this is helping eocnomic growth. However, even despite this growth, the eocnomy has not bounce dback as much as President Trump wanted it to when he took the presidency. Retail ales are also a bit disappointing this quarter. 

* Q60 --- It seems like the eocnomic numbers are verymixed. But, I still don't think Trump can take any credit for the positive things that came out of it as his policies have not really been implemented. 

* Q61 --- GDP slightly increased last quarter. Trump is struggling to lift the economy as fast as he would have liked. 

* Q62 --- GDP doesn't really mean anything.

* Q183 --- The econ is getting better and rising slowly.

* Q184 --- The economy did better than was projected. We had a 1.4\% increase when a 1.2\% increase was expected. Trump wants a 3\% increase which hasn't happened since 1990. With his tax cuts and infrastructure spending, he might get there.

* Q185 --- I'd say Trump is being effective with his plan but it needs more time to fully work out the way he intends it to work.

* Q186 --- The economy is doing very well and the stock market is going through the roof.

* Q187 --- GDP for the first quarter was revised upwards, coming in better than previously reported. That was due to greater consumer spending and exports than previously believed. However, growth is still relatively low and the economic moves Trump intended to make in regulations, taxes, and elsewehre, haven't happened yet, and aren't yet being reflected in the economy, even in anticipation. So we'll see whether the U.S. economy can ever get back to growth higher than 2\%.

* Q188 --- It appears like the economy did better than expected. However it was not by much. Still in the category of the slowest growth in quite some time.

* Q189 --- The economy is on the rise as President Trump stated it would be but it is falling short of his promised 3\%.

* Q190 --- It seems that consumer confidence has been up since Trump won. And that's even with the Democrats trying to dampen economic growth by slowing Trump down in his tax plan. Imagine how it will be onece it is passed.

* Q191 --- We are getting off to a slow start evonomically it seems.i remember Trump said he would have better than this.i am not discouraged yet though.i figure it will take time and it seems like we do have a rpesident that has a Senate working against him anyway.

* Q192 -- I believe the eocnomy will turn around greatly. I believe our goal of 3.0 percent is viable. It may take a little time, but with this administration I believe our economy will grow rapidly. 

* Q193 --- So far there have not been very much change in a good way since Donald Trump became presidnet. Companies have spend more and the numbers from the article do not look good. 

* Q194 --- Trump is making progress on improving the economy. Getting the economy he wants iwll take awhile, but he is doing the right thing. I believe Trump will be great for America's economy. 

* Q195 --- The economy is projected to grow at 1.4\% instead of 1.2\% in the first quarter of the year. We haven't had "high growth" for quite some time, and I don't imagine that we will in future.

* Q196 --- The economy is very stagnant. There are things that are going on with growth that have been similar to years past. The GDP is moving like a turtle. If Trump is to do what he said he would, he needs to get his programs running so that we can see the 3 percent growth in GDP he promised. 

* Q197 --- Trump thinks he can single handedly bring the economy back and even when it doesn't happen he still says it's so. I can't trust someone that egotistical and deluded to do anything for me, the country or the economy. Plus,I have always thought statistics were basically skewed to say whatever the person or entity wanted them to say. We are in deep trouble. It started years and a couple of administrations ago adn is continuing on, no matter what the statistics or Trump say. 

* Q198 --- GDP growth is at 1.4 percent.

* Q199 --- According to the article so far Trump hasn't done much to improve the economy. When will all of this promises be fulfilled? Do you think he is qualified to accomplish all he claimed he would?

* Q200 --- Donald Trump is very ambitious about the US growth but he sure has some challenges. It' sbeen five months and not much to show for.

* Q201 --- The economy grew at a slightly faster rate than expected last quarter but it is still rather slow. It is going to be hard to get back to 3\% growth.

* Q202 --- I read this article that was trying to portray President Trump in a poor light. It stated tha the has yet to get his financial plan off the ground 5 months into his presidency. It is so ridiculous how th emedi alike to spin everything instead of reporting unbiased facts. President Trump is much more financially responsible than the last several presidents. 

* Q203 --- Our economy has come to a crawl, but it is growing. Trump is still trying to get his tax cuts and the like off the ground.

* Q204 --- The economy has gotten slightly better since Trump became president, but it hasn't improved as drastically as was predicted. 

* Q205 --- It seems that our economy is slowly getting back to a good place despite having so many issues and roadblocks by the Democrats. Even though it's slow going, and not as fast as PResident Trump predicted, it is still going in a postive direction. 

* Q206 --- GDP rose a little more than expected, 1.4\% instead of the predicted 1.2\%.

* Q207 --- The GDp of the United states is growing and in the first quarter itself it has shown a rise from 1.2 to 1.4\% as comapred to last year. The revenue and the inventory are improving too. 

* Q208 --- The economy is doing slightly better in the most recent quarter than it was previously, but it's going to be difficult to reach Trump's goal of 3\% growth. His economic program is moving slowly, and growth reports have been mixed with increases and decreases. 

* Q209 --- The promise of an important decrease in regulatory and tax burdens on U.S. businesses suggests that the U.S. economy will grow faster than most analysts expected. In addition, and mostly due to internal rather than external factors, the U.S. economy will be a key driving force of other Western economies. 

* Q210 --- I think it's still too early to say whether or not Trump will make an impact on our economy o rnot. He has so much on his plate with so many issues I wouldn't make a hard-pressed decision on what I think of him as a President. 

* Q211 --- The article reveals that the standard economic statistics are doing slightly better than expcted, but that the economy, judged be these standards is still not back to where it was some years ago. It is not clear how well Turmp's program is doing with regards to the economy. Personally I am not sure how much thesen umbers reflect underlying reality. 

* Q212 --- You need to get your act together and support President Trump. Trump is trying to have our taxes, regulations, and infrastructure spending cut. President Trump is trying to get our country in a better place than what Obama left us in. 

* Q213 --- I read an article that basically said the eocnomy is still growing, albeit slower than before. The 3\% increase the Trump administration has been aiming for hasn't been seen since the 90's, but it has still been steadily climbing since around the year 2000. 

* Q214 --- The economic growth has not risen as expected. Trumps plans have all not taken effect yet after 5 months into his presidency. 

* Q215 --- I think the economy and the state of the union is kind of up in the air. People are trying to get a handle on where we are heading. I do feel positive about the direction we're going and if we continue on this path it will be positive for all of us. 

* Q216 --- The economy is showing signs of doing better. GDP is growing at a better rate than previously reported. 

* Q217 --- Basically Trump is growing America with the economy and jobs which trickle into many other thing like safety and security.

* Q218 --- Our economy is getting weaker everyday. President Trump has offer false hope to everyone who voted for him. He has failed as a president. You can' tpromise to do something and not do it. It is his fault tha tour economy is dwindling.

* Q219 --- The GDP was rising but slower than it had in the past.

## Social Republican Treatment --- Democrat Participants

Participants who identify as Democrats who were randomly assigned to read a message written by a Republican. Thus, these messages were written by Republicans to Democrats. All messages started with "Please read the following information that comes from a Republican who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q53 --- I think both parties need to work together to fix the economy. The stats in the article indicate that growth is slow or next to none, and that there needs to be change.

* Q54 --- The article stated that we are doing better than what originally was predicted in terms of the national economy.

* Q55 --- It will be interesting to see if the Trump administration can pull off 3\% GDP growth since the economy has grown 2\% a year roughly for the past decade.

* Q56 --- GDP has sown some signs of growing but we still have a ways to go to get to good growth of 3\%. Hope we can get there b/c it will help everyone.

* Q57 --- There appears to be significant discrepancies with the data that is being reported vs. what is actually happening. Spending and savings are on a rollercoaster with unprecedented highs and lows. There appears to be some concern that the economy is having a hard time boncingback. Thoughts?

* Q220 --- Even though the economy is slowing given time it will pick back up. Just give some time for Trump's initiatives to take affect. The article states a lot of slowing down but it does state that those initiatives have not begun as yet.

* Q221 --- Despite the fact that the President's plan to speed up economic growth has not been quite as fast as had been predicted, it does seem that consumer psending is rising a little and I think that is because of consumer optimism that the plan is working. Even though slower than expected. There also seems to be some disappointing news in regard to manufacturing production but these things take time and I believe will occur in the long run. 

* Q222 --- The economy needs to improve more to make things better. We need a boast in speanding and consumer confidence.

* Q223 --- I just read an article abotu the current economy. The economy isn't great yet but is better than expected. It certainly will be nice when it gets back to where it was in the nineties.

* Q224 --- GDP is rising at a 1.4\% clip. The current administration wants to see it grow at 3\%. The USA hasn't seen that kind of growth since the 1990's and the Internet bubble. Median incomes continue to go down and the consumer as to borrow to purchase products. I don't see how GDP can expand with wages continually falling. 

* Q225 --- I've heard that Donald Trump has actually had an increase in GDP growth. It looks like businesses are on the right track in gaining profits. What are your thoughts on that?

* Q226 --- Trump is good for the economy, things are on the up. 

* Q227 --- We need to increase corporate profit. 

* Q228 --- In the first quarter of this year, consumer spending and exports performed much better than expected. The country is on the right track under Trump and this can be seen in the optimism shown by both consumers and businesses. People believe in President Trump and are more optimistic about the country and their lives since he took office. While a sustained growth of 3\% hasn't been seen since the 1990s, that is one of President Trump's goals and the people are behind him and believe it's possibel for him to deliver because of his stated goals of rolling back regulations that were harming businesses and implementing tax cuts to help both businesses and people. 

* Q229 --- This president is not performing as promised.

* Q230 --- I acknowledge the relative strength of the economy so far during the Trump administration.

* Q231 --- The first quarter growth in the US is higher than expected, but still very slow. Newly instituted programs have yet to show any real changes in our economic growth.

* Q232 --- Love or hate Trump, you can't argue with the performance of the economy since he's taken over. The GDP doesn't lie and it's way up as is consumer spending. I don't like the guy, but Republican/conservative methods are proven.

* Q233 --- The economy is doing resonably well. there is some weakness in retail and manufacturing, but that has a lot to do with long term secular (vs cyclical shifts) in the economy. the improvement qtr over qtr is minor but things are basically moving along as they have been over the past few years.

* Q234 --- GDP is better than last month and is trending upward. We should look forward to increased economic growth and the great times that await.

* Q235 --- The economy is going to improve in a lot of way going forward. the first quarter of this year is just a starting point. if congress would get their act together and work with the president, the economy will get so much better in the future.

* Q236 --- President Trump is already doing a good job helping the economy improve. While it still has a long way to go to get to a 3\% growth, the economy actually did better in Q1 of 2017 than analysts first determined. 

* Q237 --- The Growth is a bit slower than what we had imagined, but still better than most recently. Businesses have started to pick up their inventory, and that is a great sign of things to come. I think once we revise the tax rates for all and start rolling out an infrastructure plan things will really change for the better. 

* Q238 --- Trump has made some difference. He has moved slower than he wanted due to the Democrats not supporting him. The growth has been slow but it is still there. It will grow more in time. If the Democrats would suppor thim more I see big, good changes for this country. 

* Q239 --- The economy handed to Trump by Obama was in poor shape. Trump and his cabinet are doing a lot to stimulate the economy but it will take long to show fruit. Blaming Trump for the current economy is very unfair since he's only been in office a very sthort time but Democrats will say anything to drag him down. Whether Trump's economic policy is a success will show in the next few years.

* Q240 --- GDP growth is up, consumer spending is up.

* Q241 --- Economic growth is still stagnant. Trump has not been effective at getting his changes done, but Obama didnt get much growth either.

* Q242 --- The economy has been doing so much better since the election of Trump because the stock market has faith in our President for the first time in 8 years. The main stream media does not report on the truth of the matter because they wish the public to not see how well President Trump is doing for our country. They prefer to lie and make people think Trump is failing. The economy,job market, has all improved and it is all because of Trump. If he had lost things would not have improved a bit.

* Q243 --- The economy is growing at a steady rate with a positive outlook into the future, but not without the usual bumps along the way as a countryies GDP can fluctuate at random.

* Q244 --- Trump is trying to improve our economy.

* Q245 --- While the economy still isn't back up to par, it seems to be moving a slightly better rate that was originally forcasted for this quarter. However, the government still has a long way to go to reach their preset goals.

* Q246 --- According to all reports - all aspects of the economy have started to rise.

* Q247 --- Things are turning around they have some good signs that we are headed in the right direction. Trump hasn't even got down to the things he really wants to do and by the time he gets to it people will have more faith in what he can do and what he means what he say's he just doesn't talk the talk he is willing to walk the walk. It's going to take time to clear up the mess we have been in for 8 years it doesn't happen over night. It took 8 years to get here.

* Q248 --- The president and his administration are hard at work to get the economy growing. with an increase of 1.4 percent they beat the expectations that was forcast to grow by 1.2 percent in the first quarter.

* Q249 --- We need to boost GDP and a way to do that is cut taxes, or risk a recession.

* Q250 --- It seems like the economy is slowly gaining ground, but it will be interesting to see how the new policies that President Trump is trying to enact will affect the economy.

* Q251 --- The economy is slightly improving under Donald Trump but his article proves that Donald Trump has been unable to get a grip on how American politics work. 

* Q252 --- Growth of the economy has indeed slowed, but not by as much as had first been predicted. Despite that, though, growth rate in general is slower than the previous year and expected growth is unchanged. Also, the President's goal of 3\% growth still seems unreachable.

* Q253 --- You can't always expect a consistent growth rate all of the time. There are going ot be slow times so don't be so worried when the economy slows down. The government should try to stay out of influencing the economy so that it might improve. 

* Q254 --- The US is seeing some growth, but it is still slow and will hopefully get better. The government did not have the growth they were hoping for. 

## Social Democrat Treatment --- Independent Participants

Participants who identify as Independents who were randomly assigned to read a message written by a Democrat. Thus, these messages were written by Democrats to Independents. All messages started with "Please read the following information that comes from a Democrat who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q48 --- It looks like the economy is doing a little better than expected, but less well than hoped. I guess maybe it'll continue to go up?

* Q49 --- You should read the pros and cons. Make an informed decision using your interpretation of what you understand in the article. Everyone has different takes on the subject.

* Q50 --- The article makes assumptions that econ growth is good rather than questioning the basixcc flaws of capitalist consumption leading to environmental degradation and income inequality.

* Q51 --- The economy is growing, but the rate it's growing is slower than economists would like. They readjusted the numbers on consumer spending, GDP, and similar factors to show that the economy is doing better than we thought it was doing a few months ago, but the rate is still relatively slow. 

* Q52 --- GDP went up higher than expected the first quarter but it's never very high in that quarter. The current administration promised a 3\% GDP growth which hasn't been seen since the 90s and which the current GDP fell short. 

* Q255 --- The economy is still not showing any signs of improvement under Trump. None of Trumps proposals for tax reform have gained any sort of traction. Trump wants to get the economy to numbers not seen since the 1990s and many economists say there is little to no chance of that happening and the numbers back it up. So far, the economy has continued a slight downward slope. 

* Q256 --- The economy is not grwoing as fast ot should be.

* Q257 --- Gross domestic product increased at a 1.4 percent annual rate instead of the 1.2 percent pace reported last month, the Commerce Department said in its final assessment on Thursday.

* Q258 --- The US economy has not slowed down as much as expected in the first quarter of this year. Although growth in many areas is less than expected, overall this is a positive sign. President Trump has not really achieved his primised 3\% growth yet, but this type of growth hasn't been seen since the 1990s. We usually expect about 2\% growth.

* Q259 --- The GDP incerased at a slightly higher rate than the pace reported last month, since consumers were spending more, and we were exporting more goods. It's clearly below Trump's target, because his economic program has gone nowhere. 

* Q260 --- The economy is doing well. 

* Q261 --- I bring up the point in the article where it states, that President Donald Trump's plans to help the economy has yet to provide any substantial help since he has taken office. I would also mention that the US economy has not slowed as much as predicted because of higher consumer spending. 

* Q262 --- The GDP increased to 1.4 percent annually. It is a slow growth rate. Experts are unsure if it will increase in the future.

* Q263 --- Although the GDP rose the article puts the information in a negative light. The entire article is written with a negative tone despite gains in many key areas.

* Q264 --- The economy is growing slightly but at a slower rate than predicated and about half the rate of 1990s growth.

* Q265 --- The US economy grew at a revised rate of 1/4\% in the first quarter of the year, well below the Trump administration's projections for the time period. THe administration's projections of 3\% growth seem foolhardy and doomed to fail. That rate of economic growth hasn't been seen since the 1990s. Other economic indicators were mixed. 

* Q266 --- Growth is still slow. The economy is limping along.

* Q267 --- Economic growth has been below normal. The economic changes promised by the Trump administration have yet to put forward in any meaningful way. 

* Q268 --- Our economy has been slowly grinding higher even though the policies that Trump plans to bring into action have not materialized yet.

* Q269 --- The U.S. economy slowed less sharply in the first quarter than initially estimated due to unexpectedly higher consumer spending and a bigger jump in exports. Gross domestic product increased at a 1.4 percent annual rate isntead of the 1.2 percent pace reported last month, the Commerce Department said in its final assessment on Thursday. 

* Q270 --- The economy is not growing as fast as expected. Its going to be an uphill climb to get the economy to grow by 3\%.

* Q271 --- The economy is still stagnate despite efforts to jump start it. Typically, politicians claim that deregulation and tax-cuts will create jobs but the numbers don't lie. 

* Q272 --- The U.S. economy isn't doing well or badly right now. Consumer spending has slowed, GDP grew at a rate of 1.4\% instead of a lower number which was estimated, but did not hit the 3\% that this administration has been pushing for. For the past 15 or or so years, the American economy has generally increased at a rate of 2\%. Last year it increased at a rate of 1.6\% which is the slowest it has ever been. So although the economy is increasing at the moment, it still isn't doing that well because it is increasing more slowly than it has in the past. 

* Q273 --- The first quarter of this year the GDP grew only 1.4\% though it wasn't as slow as had been predicted, it was still the slowest growth in GDP since early last year. Even though the GDP grew more than predicted it is a long way off from President Trump's promised 3\% growth. 

* Q274 --- Most of this is carryover from the Obama administration, we're still too early to tell what sort of impact Trump's economic policies will have.

* Q275 --- So the US economy is not as bad off as first thought. It's close but grew at a slightly better rate then. It's under the presidential estimate however. Also hey did you know consumer spending is 2/3's of the economy?

* Q276 --- The US first quarter GDP was revised from 1.2 percent to 1.4 percent. It still remains at the slowest growth rate.

* Q277 --- The president is putting through more policies to mess up the US. He probably has personal motive for every action he takes whether it's making money or a personal opinion about things "he" doesn't like about the country.

* Q278 --- Well the economy isn't doing all that well. Rather than the big increases Trump promised growth is still pretty slow. 

* Q279 --- The economy is eventually going to go straight to - - - - because Donald Trump who is a complete total bafoon is our president.

* Q280 --- The economy's growth is measureable, but not growing very fast, and Trump is being very deceptive and unrealistic in his claims that he can increase growth to the levels that he has stated.

* Q281 --- The economy is on the rise and it is due to a republican president and he promises more growth upwards to 3 percent currently up 1.1.

* Q282 --- While the economy has improved somewhat, it is miniscule and not near Trump's targets.

* Q283 --- By going through the info, the economy is at a challenge right now. How would you think that could get fixed?

* Q284 --- Things economically are rising in the US.

* Q285 --- This article says that the economy was growing twice as well as had been thought. Instead of about half a percentage point it was over a percentage point. Still far short of what Trump promised of three percent. 

* Q286 --- I just read this article about economic growth for the first quarter of this year. Well, it looks like Trump is doing all he promised. His tax cuts and breaks have not taken off and the economy has not grown this slowly in a long time. However, the economy did grow a bit more than expected but it is still lagging behind the growth of the past.

* Q287 --- GDP came in at 1.4\% which was slower than expected growth rate. The president's administration is promising 3\% GDP which their economic policies. The Jan thru March data is irrelevant due to calculation error by the government. 

* Q288 --- Apparently gdp went up more than expected in the first quarter, mainly due to increases in consumer spending. But we could have a discussion on whether consumer spending and gdp growth is really what we're looking for. I wonder about the environmental impacts of that. Other indicators were a little more or less than expected. 

* Q289 --- Apparently the national growth average is pretty low this year so far-not even as high as it was during the 1990s. I'm kind of not surprised to hear this based on how other things have been the last few years. 

* Q290 --- The economy is slowly growing. The tax cuts will help this growth. 

* Q291 --- Donald Trump doesn't care at all about the middle or working class. He is out to make the rich even richer and poor poorer.

* Q292 --- The trump administration has not made any significant progress towards raising the gdp that they promised. 

* Q293 --- While growth is higher than expected, it is still at a low in general, and that plans to improve this growth have not been implemented yet.

* Q294 --- Economic growth is steady in the US.

* Q295 --- The article basically stated that although the projected slow down of the economy was not as sharp as anticipated, it still was not great compared to previous records. exports helped but not by much. Trump's budget cuts didn't help boost the economy by the promised and unrealistic 3 percent. Gross domestic product jumped slightly. but the growth rate was slow. consumer spending was better than expected. 

* Q296 --- The U.S. GDP went up 1.4\% instead of the expected 1.2\%.

* Q297 --- The first quarter was good but not as good as it looks. Things are not growing like they appear to be.

* Q298 --- Trump's plan to boost the economy with a 3\% growth in GDP was unsuccessful. However, the economy was growing at a faster rate than expected.

* Q299 --- Growth is happening but at a very slow pace and that is unlikely that the Trump administration's promise of 3\% growth will actually be able to occur.

* Q300 --- I think the economy did not grow as great as it supposed to be. Obama did a better job trying to build the economy this past few years and I certainly hope Trump did not destroy it in one year. I haven't seen anything that can make US great again. 

* Q301 --- The revised GDP growth is up.

* Q302 --- All indications are the economy is strong and growing. Despite the problems the government has in tracking economic indicators in the first quarter (and they're doing things to try to improve this quarters data and reportage), they revised almost al indicators up across the board.

* Q303 --- While raising the profits from businesses is a goal with the Trump administration, it has proven to be challenging.

* Q304 --- Our president needs to do a way better job. Stuff needs to get done. Our economy should be going up.

* Q305 --- The economy has not performed as expected. Trump's efforts have yet to show any results, and while in some case things are good, such as consumption, other areas fall short of expectations.

* Q306 --- The US economy picked up a little in the first quarter, raising 1.4\% instead of 1.2\%. President Trump has called for the US to grow at 3\%, a feat that has not happened since the 1990's. Most economist think the US will grow around 1.5\% for a while.

* Q307 --- The economy is complex, it's not growing as much as we thought it would, but it is slowly improving. I don't know much about the economy, what I just read was mostly gibberish to me personally. It just goes right over my head. 

* Q308 --- This article is just an attempt to make it look like Trump is almost accomplishing one of his convoluted campaign promises. It states he's raised the GDP, but fails to mention how he's lost us billions despite this 1\% raise in GDP. We have produced slightly more, but we've lost money as a whole due to his horrendous racist, sexist, humiliating rambling he calls his foreign policy.

* Q309 --- The federal government is revising economic growth and health numbers to make the 1st quarter of 2017 appear better than previously stated. 

* Q310 --- Apparently, the GDP rose a little bit more than was originally forecast. Consumer spending was a big driver, and it was higher than expected. Growth remains lower than we've seen in the 1990s and 2000s, however.

* Q311 --- The economy is slowly less sharply than before. There isn't much growth and this trend seems as if it will remain unchanged. There is hope that policy changes such as tax changes, infrastructure changes and regulatory changes will be pu tin effect to improve the economy. 

* Q312 --- The economy has remained roughly the same since the new administration has taken over. While it has not slowed as much as predicted, it also has not gained as much as promised. There are a wide variety of factors that go into determining the overall growth such as exports, government spending, and sales to name a few. 

* Q313 --- The expected and promised growth of the US economy has yet to materialize.

* Q314 --- The US economy is shifting in an interesting direction. It slowed less sharply than predicted due to more consumer spending and greater US exports.

* Q315 --- I would ask you if you read the current GDP article about how our economical rates have been dropping after starting off with a hot start but recently under Trump have been falling and how you feel about how Trump is currently managing the Economy and what you think about the falling rates.

* Q316 --- The national GDP has actually risen more than we thought it has during the first quarter under Trump.

* Q317 --- The US economy has started to grow at a faster rate than the previous years and even previous quarter. The recent presidency appears to have had an impact in the uptick and the economic market has not slowed but expanded and is expanding. 

* Q318 --- You need to start leaning to one side of the spectrum, at least on very specific issues. This article I read clearly indicates that the US economy is showing a nearly standstill growth. This is largely due to the do-nothign Republicans in the house \& senate. The Democratic way is the only way to progress this country forward. The growth rate is the slowest it has been in 5 years. Seems like the Obama administration was doing something right, eh?

* Q319 --- It looks like the economy has seen a boost in the second quarter, but it is less strong than both anticipated and Trump's promises. Retail and manufacturing are down, while consumer spending is up. The goal is 3\% increase, but it's 1.7\%.

* Q320 --- The economic growth in US in first quarte rof this year was higher then expected but still way lower than 3\%. 3 \% is what new President said will happen. Last time we saw numbers around 3\% was in 1990.

* Q321 --- Our main goal is to advance our country and communitites to have a better life for their amilies and not have to work like slaves. Under the current administration, our sales are dropping our cousin living is rising, we are feeling and many parts in the and ministration that we need to advance our country. Logically the right or the left isn't thinking clearly. Our president needs to step down and quit lying to us about these ratings that are false, bigoted, and delusional. Number 45 said he would bring jobs back to America and he hasn't even put his pinky on it, he's more worried about building a wall and spitting in the face of our transgender Military men and women that have given their lives for us. I don't care who you are, for the love of God there needs to be an impeachment.

* Q322 --- Economic activity in the first quarter, annualized, was 1.2\%, which was higher than expected due to an increase in consumer spending and exports. Donald Trump is aiming for 3\% annualized economic growth, which hasn't been seen since the 1990s. It will be very hard to achieve 3\% annualized economic growth. 

* Q323 --- I read that us gross domestic product had increased in the last few months. Eventhough there was initial increase in economy under Trumps administration now it has been suffering to keep up with the expected growth. 

* Q324 --- I have doubts that Trump's economic forecast will come to pass. Our economy needs a boost, but I don't see a plan that will raise it by 3\%.

* Q325 --- This government has not done anything it had promised. The economy is slowing down and even after tax cuts the spending has not gone up. Last year was much better. The growth rate has decreased. 

* Q326 --- Experts are saying the first quarter of this year actually did better than they forecasted. However, Trump has yet to get started on some of the things that could majorly affect our economy, so we have to see how that goes. 

* Q327 --- The economy is slowing down.

* Q328 --- An article I just read just stated that the US economy grew at the slowest rate since last year. It is growing slightly faster than originally estimated, at 1.4\% in this 2nd quarter, but that is still pretty far from Trump's goal of 3\%. A lot of the programs he proposed to get there have yet to become a reality. Also exports and businesses did pretty well last quarter. Consumer is up a little bit as well. 

## Social Democrat Treatment --- Republican Participants

Participants who identify as Republicans who were randomly assigned to read a message written by a Democrat. Thus, these messages were written by Democrats to Republicans. All messages started with "Please read the following information that comes from a Democrat who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q41 --- GDP growth has been much less than expcted, and even further less than that promised by President Trumps proposed policy changes.

* Q43 --- My understanding is that your party being in office doesn't seem to have fixed anything. According to a report I just read GDP isn't growing as fast as predicted, nor is the current administration's plan taking off.

* Q45 --- Well, the numbers seem to show that Trump's economic programs are having little if any impact on the actual economy. Growth is at a slower rate than it has been since 2013. The promised infrastructure spending looks like it is not happening, and I doubt it will unless they raise taxes. Reading this report on the actual statistics does not show much progress towards his promised economic growth. Not surprising considering his actions I would say.

* Q46 --- Our domestic product income has increases from 1.2 to 1.4 percent. This is good but still pretty low. Trump has nothing to do with this though as his plan has yet to take off. 

* Q47 --- I feel the presidents agenda is not being implemented. Nothing he promised is happening because all there is is constant controversy. 

* Q329 --- Promises made by the president during his campaign have not come through at all. Not one bit of legislation has gone through the Republican legislature. Obviously republicans are great and complaining and broken promises but are a do nothing party when in power.

* Q330 --- The GDP increased at a 1.4 percent rate. Which is slower than it has been in a long time. Most of Trump's economic policies have yet to take off.

* Q331 --- GDP growth was still relatively low for the first quarter of Trump's presidency.

* Q332 --- Thanks to our idiot orange haired president, our economy will grow slower rather than quickly.

* Q333 --- Data is only half the story.  It's easy to cook the numbers without taking into account the sociological factors.  From  a citizens standpoint growth does not feel like it is real.  I question the methodology and the accuracy of economic grown claims.

* Q334 --- Hey, I didn't realize the Republican party was doing so well this term!

* Q336 --- Economic growth has yet to occur under the Presidency of Donald Trump. He promised reform, so where is it?

* Q337 --- The U.S. economy has slowed but not as sharply due to a jump in exports. It's a slow growth but at least it wasn't as bad as was originally thought.

* Q338 --- Trump's goal of a 3 percent GDP gain annually is unrealistic and hasn't been seen in nearly 20 years. The backbone of the economy is consumer consumption and Trump should focus on ensuring that wages rise so that can continue.  It is too early to see any effects from a Trump presidency.

* Q339 --- The Trump administration has done little to deliver its promises of higher GDP growth. His promise of 3% growth has not come to fruition. More needs to be done.

* Q340 --- Economic spending is on the rise, slowly, but still rising.

* Q341 --- Our economy this quarter increased by 1.4 and it was predicted to be 1.2. It is still a slow growth compare last year but it still a good news.

* Q342 --- GDP growth has been the slowest since the second quarter of last year. Economic growth is also at the slowest pace since the second quarter of 2013. The new presidency is bad for the economy so far.

* Q343 --- This article puts a lie to the Trump administration's claim that the nation's GDP would grow to 3\%. It is simply not going to happen. As the article points out, GDP has not been that high since the 1990's. Republicans and their presidents have mismanaged the economy so badly, that Democratic presidents like Barack Obama have had to come in and clean up their mess.

* Q344 --- The economy is stale. It is not growing at the rate it needs to grow. There needs to be ore movement. 

* Q345 --- Trump seems adamant that the market will grow, even though the downturn has been sharp. I just don't think he likes to face reality.

* Q346 --- The economy has not improved since Trump took office.  It showed some growth at the beginning of the year but that would be due to actions taken during Obama's term. 

* Q347 --- Honestly, it was not easy to read the article I just read. At first, I thought the article was saying that there was an increase in the economy due to consumer purchasing and exports, but further down the article, it started to state that other consumer spending was at a low and corporate profits had dropped. So, I guess the economy is growing in some areas and suffering in others.

* Q348 --- There article threw a lot of values at me but didn't tell me what they meant to it is perfect for republicans who just make up their own - - - - anyway.  Apparently allot is higher then projected but it isn't good because profits are not but I don't know why as the article didn't tell me anything about what anything meant.

* Q349 --- The US economy is slowing because of the man you voted for, Donald Trump.

* Q350 --- The revised GDP numbers do not reflect a good rate.

* Q351 --- I saw that they revised first quarter GDP growth up a couple of ticks.

* Q352 --- The economy is doing better than expected. I think it's because the expectations of tax cuts that will drive spending.

* Q353 --- Despite Trump's promise on behalf of Republicans that he is go to make America great again, he hasn't done much to help our economy.

* Q354 --- Seems like your boy Trump isn't living up to all of his promises, huh?  We're not at that 3% he promised and it's not looking likely that we will be.

* Q355 --- What the president said about the economy was wrong, and he hasn't done anything yet to help the economy. Instead he has been embroiled in controversy after controversy, and his inabilty to control himself for one thing is very unnerving.  He has no previous experience in government and it shows. He is worthless in my opinion.

* Q356 --- The US economy did slightly better than expected due to people buying more things, but it's only a slight increase. Trump's plan to increase the economy to 3\% is a tough hurdle.

* Q357 --- While the the estimate was revised upward, it seems that it is still a ways away from the 3% target of the administration, and is below the historical average.  With his agenda seemingly stalled, how do you think the President is going to get the nations economy humming again?
 
* Q358 --- The economy is heading downward and needs to be addressed.

* Q359 --- While that sounds positive, I don't know if its something Trump can take credit for. It is probably left over of the Obama administration. And Trump's goal of 3\% is unrealistic considering he doesn't even have a tax proposal yet. 

* Q360 --- Trump has his work cut out for him, but it looks like we're heading in the right direction. The GDP grew more than expected, which indicates that the economy is on an upward trend. If Trump can implement those corporate tax cuts he was talking about, I think we can turn this economy around completely.

* Q361 --- The US economy slowed less in the first quarter than estimates predicted because of unexpectedly higher consumer spending and a bigger jump in exports. Our gross domestic product increased at 1.4 percent instead of 1.2 percent. But this is just momentum from the previous administration. Trump has had no impact on any of it.

* Q362 --- Trump expecting growth of 3% has happened since the 1990's, and he hasn't done anything yet to help that.

* Q363 --- Our economy is doing fairly well. We are growing as a country and we are selling more products. The growth will help with jobs and the economy.

* Q364 --- The Trump administration has still not been able to get going their plan to raise the U.S. growth by 3\% and it's already been 6 months. 

* Q365 --- I have recently read an article about the economy of US after trump presidency. It is not that much satisfactory. The growth is very less as compared to first quarter. All because of presidents economic program of tax cuts, regulatory rollbacks and infrastructure spending. It is not good for US economy. We do not feel any good growth in his 5 months of presidency.

* Q366 --- The Trump agenda should lead to decreased regulations and business tax cuts, which Republicans believe will propel the economy to new highs.  The stock market has spiked to new highs while investors await word that infrastructure spending and tax cuts have been moved through the congress and are waiting for the President's signature. Hopefully, healthcare will get through under reconciliation and allow for the agenda to move forward.  Investors remain hopeful despite the stalls coming from inside the Belt Way.

* Q367 --- The Trump Administration, and fellow Republicans have failed to improve the economy, so to speak. He expects to strengthen the economy by 3%, but has failed to reach that.  According to the Article, GDP growth has only reached a staggering 1.4%. The regulations rollbacks, tax cuts for corporations, and so forth have not worked, since the five months he has been office. Trickle down economics apparently is not working. 

* Q368 --- It's shocking that the percentage is higher than expected, but it's still tough that overall it is a low number. The president is pushing his luck wanting 3 percent.

* Q369 --- The U.S. economy slowed less in the first quarter due to unexpectedly higher consumer spending and a bigger jump in exports. Consumer spending rose although it was still the slowest pace since the second quarter of 2013. 

* Q370 --- It would appear that in the new President's first five months in office he has not been able to fulfill his promises to booth economic growth in the United States. He hasn't been able to implement successfully his economic program, and as a result, the country's economic growth is stunted. What do you believe needs to be done to stimulate economic growth? Is it your opinion that the President and his administration will be able to formulate and successfully enact a plan?

* Q371 --- Instead of the 1.2 percent increase that usually occurs, there was a 1.4 increase this past month. 

* Q372 --- The GDP rose more than expected at 1.4 percent this past quarter compared to 1.2 percent the quarter before. However, this is still less than economists would like to see. President Trumps plan to increase the GDP to 3 percent has still not gotten off the ground 5 months into his presidency.
 
* Q373 --- I just read an economic article which states that the GDP growth in the first quarter was actually higher than initial estimates projected.  Consumer spending was higher than predicted and this could be a result of the president's policies although they are still getting off the ground.  It's not as good as it was in the 90's but still I think the economy is heading in the right direction.

* Q374 --- Donald Trump is inaccurate in his GDP claims. There has been a small amount of GDP growth. 

* Q375 --- It looks like our current president has been doing good things for the economy lately.

* Q376 --- There is a chance that the economy is boosted. There are some things that point to it slightly increasing.

* Q377 --- First quarter GDP growth was revised from 1.2 percent to 1.4 percent. It was a relatively slow growth period. From various indicators like consumer spending, the economy grew, but fairly slowly.
 
* Q378 --- Basically, economic growth isn't happening as quickly as Trump expected. His plan for 3\% is a bit of a lofty goal, as the current rate is about 1.2\%, the lowest in recent memory. That said, it's not impossible, as the US had reached 3\% back in the 90's, and even the Atlanta Federal Reserve is predicting an almost 3\% increase in the second quarter of the year.

* Q379 --- The economy is growing at a little over 1\%. There is more consumer spending, but the goal of 3\% growth rate seems entirely unlikely.

* Q380 --- Growth in the united states is slowing down.

* Q381 --- THE ECONOMICAL FORECAST YOUR PARTY RELEASED WAS WRONG.

* Q382 --- Trump is not doing what he said that the would do. You say that we needed someone that is not a politician but a business man, and he is still underperforming. We need change not stories and lies, or threatening tweets.  

* Q383 --- I can't believe trump is trying to hold on to unbelievable and unsubstantiated claims on the economy.  The data shows how it is responding to Obama's policies coupled with his election.  All his bluster and hot air cannot change the data.
 
* Q384 --- The economy is not improving under Trump. He has put unqualified people at all positions of government. He is also the most unqualified and disgusting president we have ever had. It will not be long before his policies and Congress policies tnk our economy completely.

* Q385 --- Economic growth isn't going too great. We're doing okay but it's nothing like the growth Trump promised.

* Q386 --- The U.S. economy slowed not as fast as expected in the first quarter than initially estimated due to higher consumer spending and a bigger jump in exports. This is great.

* Q387 --- Trump promised a 3\% growth rate which has not been since the 90's. While there has been some growth this last quarter, it's not what it needs to be to hit Trump's goal.

* Q388 --- A lot of spending and GDP were higher than expected. It seems as though in general, the economy is better than what was previously reported.That's a good sign.

* Q389 --- The article really goes to show how slowly our economy is growing annually. At a 1.4 percent, it is only sightly higher above the 1.2 normal growth usually experienced. This tells us that whatever we are doing right now does not work. Our economy needs to go back to the 90's era where the growth was at a 3 percent rate.

* Q390 --- It seems that the economy is slowly knitting itself back together, no thanks to Trump and the GOP in congress.

* Q391 --- The government released its revised economic report for the first three months of this year. It showed modest growth, with mostly minor revisions from the previously released report for that period.

## Social Democrat Treatment --- Democrat Participants

Participants who identify as Democrats who were randomly assigned to read a message written by a Democrat. Thus, these messages were written by Democrats to Democrats. All messages started with "Please read the following information that comes from a Democrat who read an article about the US economic performance in the first quarter of 2017 published by Reuters"

* Q35 --- The economy is stalled and Trump's economic ideas aren't getting passed. 

* Q36 --- Trump is going to have a tough time getting to his goal of 3 percent growth. 

* Q37 --- The GDP is up a little but not what they were expecting considering all of the promises made. 

* Q39 --- The article discussed the GDP. The economy slowed less sharply than estimated because of consumer spending. However it was still a slow growth rate.

* Q40 --- The growth is slightly above the projection, but nowhere near what Trump had said. 

* Q392 --- GDP rose a bit higher this quarter than people expected, 1.4\% instead of 1.2\%. Still disappointing overall, though.

* Q393 --- The economy has slowed more than was expected in the past quarter.  The government's projections are probably off by about half. 

* Q394 --- Our economy isn't doing as well as we hoped it would under the new administration.

* Q395 --- The economy is growing at a sluggish pace, it's nothing fantastic but at the same time it is growing. Considering the law decade, this might be the new normal, the years of 3% economic growth seem to be a thing of the past. Trump hasn't delivered a single one of his economic promises so far.

* Q396 --- The US economy has done quite successfully in the first quarter of this year. It slightly surpassed the projected growth.

* Q397 --- It seems fairly obvious that for all of Trumps promises, the economy is continuing to stagnate, and there is very little growth happening.

* Q398 --- It seems that the spending rate isn't falling as quickly as had been projected. Trumps plan to increase the economy at a 3\% rate was ambitious to say the least and make note of the fact that his economic plans haven't gotten off the ground even five months into his presidency.

* Q399 --- Despite being a horrible human being, the country has prospered somewhat since Trump's getting into office.  Most likely just because of conservatives trying harder because they are motivated, not necessarily because of anything he has done.

* Q400 --- Democrats would think that Trump is just lying to himself and everyone about the economic growth, and that we will see even more decrease soon.

* Q401 --- The United State's economic growth is lower than what was forecasted. Last year's second quarter was the lowest quarter of the year and although the President has said that we would see major gains into his presidency, this does not appear to be true.

* Q402 --- Overall, Trump's aim to increase the GDP growth to 3\% is getting off to a very slow start.  He has not been able to implement any of his tax cuts that aim to help increase growth.  The GDP growth rate is now at 1.4\% compared to 1.2\% in previous months.  There appears to be more spending by businesses and consumers, but this is not getting us to the 3\% GDP growth that Trump is aiming for.

* Q403 --- Although the current market is slowly rising, I'm not sure the market can sustain it. It's been below average, or where it should be since the late 90s. With that being said, can we really blame a president's actions. Sure, the policies accepted or veto have a direct correlation, however, it appears that the true power of the US economics lies with some other group/party. 

* Q404 --- Seems like the economy is slowly starting to come back, despite all the bull- - - - politics. 

* Q405 --- I see that we are not meeting the rates that were expected and we are being hit with losses and I don't know what to expect in this world. I don't understand everything, but I saw lose and no expectations met. I do not understand government and what is happening or why this is happening. 

* Q406 --- I just read an article in which Trump administration revise their report to make it look like our economy is doing great. Trump policy hasn't even get off the ground and they're reporting that our economy is gaining this much. 

* Q407 --- The economy slowed down but it's still at a steady rate. Things are not going down and it seems things might be starting to go well.

* Q408 --- The first quarter is getting better and things might be starting to look up.

* Q409 --- Trump is really far off the mark of 3% GDP growth he promised. GDP growth has dropped since he became President.

* Q410 --- To be fully honest I probably wouldn't even mention this article while discussing politics/current events. The connection between federal government and national economics is not so strong as to be a clear political link, especially with a new president (five months in office is not long enough for any of his policies to have taken such an effect they'd move businesses on a massive scale). If I did discuss it, it'd be to eyeroll over the administration's semi-regular attempts to point to any economic boost as "see! we're good for the economy!!!" even when referencing events which were put in motion years ago and have - - - - -all to do with government policies.

* Q411 --- Growth in the first quarter of the year was slow, though not as slow as previously expected.  It is a challenge for the administration to do what it promised and get economic growth to 3  percent.

* Q412 --- The economy hasnt been specifically going through a downfall but it certainly has been slowed down im saying in a way of consumption and imports along with exports.

* Q413 --- Looks like we should have voted with you, for Hillary Clinton.  Our man Trump is not living up to his promises.  He said 3 percent growth, but nope. 

* Q414 --- Wow, consumer spending is two-thirds of the economy. Did you know that? Donald Trump's economic program is not working after being in office for 5 months now! He sure brags about his business acumen so I hope that he can live up to his promise. I know you can't stand him, but I hope you want him to achieve his goal of boosting U.S. growth to 3 percent. He is not the person that I voted for at all. Despite that, I still want the economy to do well.

* Q415 --- I don't think there is much to say. Trump has not yet achieved his goals and it is probably too early to determine how Trump's policies will affect the economy. There isn't really much this article has to do with Trump in the first place, besides mentioning that his policies are not yet implemented.

* Q416 --- The economy isn't really growing, as expected. It looks across the board for everything to be below what was estiamted. 

* Q417 --- The economy is starting to improve slightly because more people are spending their money.

* Q418 --- It seems as though the economy is not recovering at a very quick rate. The 1990s saw the best growth years. Predictions about growth have been wrong also, we are behind what the predictions have said would happen.

* Q419 --- Economic growth is not great right now.  There is too much money being gained by big business.

* Q420 --- GDP growth has increased slightly to 1.4 % in the first quarter of 2017. This very modest increase is thought to be due to higher consumer spending and an increase in exports. It is still a far cry from the fast 3% growth promised by Trump. A sustained  GDP growth of 3 % has not been seen since the 1990s. Average GDP growth is around 2% over the last several years.

* Q421 --- It looks like the economy is doing better than previously thought. As a nation we are reporting better than we are. I would recommend evaluating what is causing the increase to the GDP and see if we can boost these items further. 

* Q422 --- It seems like the economy is doing alright, it is not growing as quick as it had, but at least it is headed in the right direction.

* Q423 --- Even though we are making a progress, the rate of economic growth is just took slow.  We need to do something to jump start the economy. The slow growth of 1.6\% is not acceptable. Democrat should outline their economic agenda that would help the country.  A lot of people are beginning to worry about the economy.  Democrat should let people know what they are going to do to make sure that we have substantial growth in out economy.

* Q424 --- There has been some very marginal improvement but not compared to previous years.

* Q425 --- GDP is growing more than they expected, but still slower than under the previous administration. 

* Q426 --- In general the first quarter of this year was a disappointment not meeting target levels set by President Trump.

* Q427 --- The economy seems to be on the upswing according to a Reuters article I read. It has nothing to do with Trump's efforts as his policies have yet to be implemented.

* Q428 --- The economy under the new Republican presidency is slowing crumbling. Nothing is being done about our currently financial situation because no one can agree on any type of final plan for the United States. I fear what will happen if rates continue to fall and growth continues to slow down. 

* Q429 --- Gdp is expected to go up.

* Q430 --- The economy is showing signs of improvement thanks mostly to increased business confidence which will spur hiring and investment. 

* Q431 --- The economy is looking ok. Growth is slowing down, but most of the fundamentals are holding steady and there is no sign of a recession as of now.

* Q432 --- The present administration has boosted the confidence in people to the point that they are relaxed in making purchases decisions and the economic future. But because our manufacturing is mostly overseas the money being spent is not reflective in our nations GDP in a positive way.

* Q433 --- Trump's promise to raise GDP to 3\% has failed at less than half that.

* Q434 --- This shows that Trumpism is bad for the economy even if there have been some localized upward trends here and there.  Things are not quite as bad as economists have predicted but they are still bad compared to where they have been in quite some time.

* Q435 --- I just read an interesting news article about the economy. It seems that they've been seeing some expansion in most sectors, albeit by just a little. I wonder if this means that corporations simply have more faith in Trump as president?

* Q436 --- GDP Growth slowed down in the last quarter.

* Q437 --- The economy is stable but does not have a lot of forward momentum.

* Q438 --- Trump has failed to kickstart the economy with his ideas and policies.  After five months the outlook is bleak despite marginal gains over expectations.

* Q439 --- The economy is gettng used to Littlefingers , it has not slowed as expected but  it is still a long way from the Cheeto's promise of 3\%.

* Q440 --- The GDP has seen growth recently. I don't remember the specifics, but the trend is upwards, albeit the 3\% figure is still an up hill battle. But it's about half way there.

* Q441 --- We've had our first opportunity to see how the economy has performed under the Trump Administration.  I think that it's really too soon to tell how much of this report is due to Trump's performance/influence on the market, and how much might be a hold-over from 2016.  I will be much more interested in seeing his 2nd quarter numbers, and then, down the road, any year-over-year numbers that emerge, as a way of really assessing what Trump's impact on the economy really is.

* Q442 --- The first quarter GDP was revised upward slightly from 1.2\% growth to 1.4\% growth. While its been revised upward, the economy grew at slowest pace since early last year. The current administration's goal of 3\% GDP growth will be hard to attain consiering it hasnt been accompished since the early 1990's.

* Q443 --- I just read an article saying that economic growth was stronger than expected in the first quarter of the year. However, it is still on the low side compared to recent years, and far short of the BS projection of the current regime in Washington. I guess it's going to take a little while for 45's administration to blunder into financial disaster for the country, but I have every confidence that they will do so, alas.  

* Q444 --- Economics is confusing and things aren't improving as well as they were said they were going to.

* Q445 --- The economy is doing better than expected per the GDP measure. Consumer spending and exports were higher than predicted. The economy's rate of expansion has been relatively low, so this may be a suggestion of that changing.

* Q446 --- The economy is still doing pretty well, but it's not because of the new President.  I'm sure he'll take the credit for exporting and job growth, but that's from the Obama Administration policies.  Job growth is a full percentage point behind where Trump claimed he'd get it to 3\%.

* Q447 --- You were right. Trump made promises about being able to grow the economy at a tremendous rate. Instead its lower than previously, with little hope for growth in the near future.

* Q448 --- Trump is doing as well at stimulating the economy as he said he would. Trump aimed for 3\% gdp growth, which hasn't happened since the 90's, but we're still sitting at 1.4\%. So, what a surprise, he's mostly just full of hot air.

* Q449 --- Our current president has not made any issues or passed any bills so far and the GDP is lower than expected, although it was higher from Jan. to march, but then dropped off.

* Q450 --- The article seemed to indicate the economy is slightly better off in some areas with growth lagging behind in others. Seriously, considering the turmoil the country has been in over the course of the past 5 months it's surprising the economy is doing as well as it is. 

* Q451 --- The U.S. economy rose 1.4 percent in the first quarter of 2017. All of President Trump's initiatives have not started yet to boost the economy. 

* Q452 --- The economy slowed less than economists expected it would, with a 1.2\% growth, which isn't great, but not as bad as expected. It was caused by consumer spending and an increase in exports. 

* Q453 --- Obama's economy continues to chug along but Trump's inaction is slowly causing damage.

* Q454 --- The economy is growing, though at a slower rate since Q2 of 2016.  The initial number on the economic growth was revised upward.  Consumer spending is slow

* Q455 --- Consumer spending went up and gdp is going well.

* Q456 --- I read that the economy isn't growing as much as it did last year. The 1st quarter was one of the slowest in recent history. 

* Q457 --- The economy continues to grow slightly better than expected, but slower than in recent quarters.  Consumer spending increase drove the higher than expected growth.  It continues to be unlikely that the annualized growth will be the 3\% that Trump had promised.

* Q458 --- Trump is the worst isn't he?  Overestimating our economy will do nothing but hurt us.

* Q459 --- Despite the earlier forecast Americas GDP has grown at 1.4 \% rather than the 1.2 \% that had been estimated.This despite the failure of President Trump's tax incentive, infrastructure spending etc that have yet to take off five months into his presidency.There have been marginal spending with businesses buying more machines, consumers spending more , auguring well for the GDP. 

* Q460 --- All the things you feared would happen with trump are happening, and despite all the promises he made, he has not come through with any of them yet, and the outlook for keeping his promises looks NOT promising.

* Q461 --- Trump's economic deregulation hasn't yet taken off the ground, but GDP growth in the first quarter is slightly higher than predicted. It's still a long way off the 3% rate of the Trump team's goal, but that goal appears to be unrealistic as rates like that haven't been around since the 90's.

* Q462 --- It seems like the economy is no longer stable. One minute the economy is doing great but then data reflects that is TRULY not!  I wish Obama was back in office. At least we saw great improvements with the economy since his early presidency. By the way this article stated the economy has been bad since 2000 I beg to differ!!

* Q463 --- The economy had only modest growth in this year's first quarter. President Trump has not put into effect most of the ideas he had to spur economic growth, such as infrastructure improvement, deregulation, etc.

* Q464 --- The economy is doing well, and things in the first quarter are up, but it doesn't seem to have anything to do with the Trump administration, despite him continually trying to take credit for it. 

* Q465 --- You have to look at the positive growth. You want more citizens to have good paying jobs and this is the fast way to do it. When companies do well they hire. Simply. This gives money to those employed.

* Q466 --- Looking at this article it is proof that the economy is growing under the current Republican administration. It is growing a bit slower than expected, but that might look to change. It's still going to be difficult to reach a 3 percent yearly growth..

* Q467 --- The government is not doing a fine job in making the economy better after all the promises that they made.

## Media Treatment 

Please read the following information about the economic performance in the US in the first quarter of 2017. This information comes from a news article published by Reuters.

U.S. first-quarter GDP growth revised up to 1.4 percent
By Lindsay Dunsmuir | WASHINGTON, JUNE 29 | Reuters

The U.S. economy slowed less sharply in the first quarter than initially estimated due to unexpectedly higher consumer spending and a bigger jump in exports.

Gross domestic product increased at a 1.4 percent annual rate instead of the 1.2 percent pace reported last month, the Commerce Department said in its final assessment on Thursday.

It was still the slowest growth rate since the second quarter of last year. Economists polled by Reuters had expected GDP growth to remain unchanged at a 1.2 percent rate.

GDP for the January-March period tends to underperform relative to the rest of the year due to perennial issues with the calculation of the data the government has said it is working to resolve.

First-quarter economic growth was boosted by an upward revision to consumer spending, which accounts for more than two-thirds of U.S. economic activity. Consumer spending rose at a 1.1 percent rate instead of the previously reported 0.6 percent pace. It was still the slowest pace since the second quarter of 2013.

Despite the upward revision, the Trump administration's stated target of swiftly boosting U.S. growth to 3 percent remains a challenge.

A sustained average of 3 percent growth has not been seen since the 1990s. Since 2000, the U.S. economy has grown at an average 2 percent rate. The U.S. economy expanded 1.6 percent in 2016, the lowest rate in five years.

President Donald Trump's economic program of tax cuts, regulatory rollbacks and infrastructure spending has yet to get off the ground five months into his presidency.

Initial signs that economic growth re-accelerated sharply in the second quarter have also faltered with recent disappointing data on retail sales, manufacturing production and inflation. Housing data has also been mixed. The Atlanta Federal Reserve currently forecasts annualized GDP growth of 2.9 percent in the second quarter.

Exports in the first quarter were revised to show a gain of 7.0 percent from the previously reported 5.8 percent.
Business spending on equipment was revised to show it increasing at a 7.8 percent rate in the January-March period rather than the 7.2 percent previously estimated.

Businesses accumulated inventories at a rate of $2.6 billion in the first quarter, rather than the $4.3 billion reported last month. Inventory investment rose at a $49.6 billion rate in the fourth quarter of last year.

Inventories subtracted 1.11 percentage point from GDP growth instead of the 1.07 percentage point previously reported.

The government also reported that corporate profits after tax with inventory valuation and capital consumption adjustments fell at an annual rate of 2.7 percent in the first quarter after rising at a 2.3 percent pace in the prior three months.
